chrissewell/whisperX
1
0
Fork 0
mirror of https://github.com/m-bain/whisperX.git synced 2025-07-01 18:17:27 -04:00
Code Issues Packages Projects Releases Wiki Activity

skeleton v2

Browse Source
This commit is contained in:
Max Bain
2023-03-30 05:31:57 +01:00
parent 1e7c2c337b
commit 18b63d46e2
53 changed files with 752 additions and 106949 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
examples/sample01.mp4
View File

Binary file not shown.

BIN
examples/sample01.wav
View File

Binary file not shown.

BIN
examples/sample_de_01.wav
View File

Binary file not shown.

BIN
examples/sample_fr_01.wav
View File

Binary file not shown.

BIN
examples/sample_it_01.wav
View File

Binary file not shown.

BIN
examples/sample_ja_01.wav
View File

Binary file not shown.

BIN
examples/whisper/sample01.mp4
View File

Binary file not shown.

140
examples/whisper/sample01.wav.srt
View File

@ -1,140 +0,0 @@
1
00:00:00,000 --> 00:00:03,000
Bella, Gloria, love.
2
00:00:03,000 --> 00:00:04,000
Oh.
3
00:00:04,000 --> 00:00:05,000
How are you?
4
00:00:05,000 --> 00:00:07,000
Oh, I'm OK.
5
00:00:07,000 --> 00:00:08,000
I will be.
6
00:00:08,000 --> 00:00:09,000
I said she could stay with us tomorrow
7
00:00:09,000 --> 00:00:10,000
just until she feels better.
8
00:00:10,000 --> 00:00:11,000
Yeah.
9
00:00:11,000 --> 00:00:12,000
Of course she can.
10
00:00:12,000 --> 00:00:14,000
No, things won't be for long.
11
00:00:14,000 --> 00:00:16,000
Well, you can stay as long as you want, my love.
12
00:00:16,000 --> 00:00:18,000
I've really missed you.
13
00:00:18,000 --> 00:00:19,000
Pops.
14
00:00:19,000 --> 00:00:20,000
Great to see you, love.
15
00:00:20,000 --> 00:00:22,000
Oh.
16
00:00:22,000 --> 00:00:23,000
All right, shall we get you off to bed then?
17
00:00:23,000 --> 00:00:25,000
You should have given me some warm.
18
00:00:25,000 --> 00:00:26,000
I know.
19
00:00:26,000 --> 00:00:27,000
I'll have to put the electric blanket on.
20
00:00:27,000 --> 00:00:28,000
I'm sorry.
21
00:00:28,000 --> 00:00:29,000
All right, Bella.
22
00:00:29,000 --> 00:00:31,000
Freezing up there.
23
00:00:31,000 --> 00:00:34,000
In a bedroom, Peter unpacks her suitcase.
24
00:00:34,000 --> 00:00:38,000
The middle-aged woman opens her green case.
25
00:00:38,000 --> 00:00:39,000
Do you want your PJs?
26
00:00:39,000 --> 00:00:40,000
Yeah.
27
00:00:40,000 --> 00:00:42,000
Yeah.
28
00:00:42,000 --> 00:00:45,000
Lifting a bundle of pajamas, Peter finds a sheet of paper
29
00:00:45,000 --> 00:00:50,000
labeled Lancaster North Hospital discharge sheet.
30
00:00:50,000 --> 00:00:52,000
He closes the suitcase and brings Gloria the pajamas.
31
00:00:52,000 --> 00:00:54,000
There you go.
32
00:00:54,000 --> 00:00:55,000
Thank you.
33
00:00:55,000 --> 00:00:57,000
He picks up the locket.
34
00:00:57,000 --> 00:00:59,000
He kept it.
35
00:00:59,000 --> 00:01:28,000
Oh, cool.

92
examples/whisper/sample_fr_01.wav.srt
View File

@ -1,92 +0,0 @@
1
00:00:00,000 --> 00:00:01,240
Lâchez, c'est bon.
2
00:00:01,240 --> 00:00:02,240
Ça va?
3
00:00:02,240 --> 00:00:03,240
Oui.
4
00:00:03,240 --> 00:00:04,240
Merci beaucoup.
5
00:00:04,240 --> 00:00:05,240
Chèque ou espèce?
6
00:00:05,240 --> 00:00:08,640
J'ai un chèque sur la commode, il est signé.
7
00:00:08,640 --> 00:00:09,640
Je vais le repirer.
8
00:00:09,640 --> 00:00:10,640
Ok.
9
00:00:10,640 --> 00:00:11,640
Ouh là!
10
00:00:11,640 --> 00:00:12,640
Venez.
11
00:00:12,640 --> 00:00:13,640
Merci.
12
00:00:13,640 --> 00:00:14,640
Ah! C'est qui?
13
00:00:14,640 --> 00:00:21,640
C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
14
00:00:21,640 --> 00:00:26,640
Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
15
00:00:26,640 --> 00:00:27,640
Ça va?
16
00:00:27,640 --> 00:00:44,200
Attendez, tout à l'heure là, c'était vous? Vous? Pas lui? Vous?
17
00:00:44,200 --> 00:00:48,360
Vous avez tout à fait raison, M. Xanaquis, Malek est à l'interne brillant qui apprend
18
00:00:48,360 --> 00:00:49,360
le métier avec moi.
19
00:00:49,360 --> 00:00:50,360
Ah!
20
00:00:50,360 --> 00:00:51,360
Bien.
21
00:00:51,360 --> 00:00:55,520
Justement, il y a la famille Boboune qui m'attend pour une consultation.
22
00:00:55,520 --> 00:00:56,520
Qui?
23
00:00:56,520 --> 00:00:57,760
Faisons pas attendre les Boboune, allez.

23
examples/whisper/sample_fr_01.wav.txt
View File

@ -1,23 +0,0 @@
Lâchez, c'est bon.
Ça va?
Oui.
Merci beaucoup.
Chèque ou espèce?
J'ai un chèque sur la commode, il est signé.
Je vais le repirer.
Ok.
Ouh là!
Venez.
Merci.
Ah! C'est qui?
C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Ça va?
Attendez, tout à l'heure là, c'était vous? Vous? Pas lui? Vous?
Vous avez tout à fait raison, M. Xanaquis, Malek est à l'interne brillant qui apprend
le métier avec moi.
Ah!
Bien.
Justement, il y a la famille Boboune qui m'attend pour une consultation.
Qui?
Faisons pas attendre les Boboune, allez.

71
examples/whisper/sample_fr_01.wav.vtt
View File

@ -1,71 +0,0 @@
WEBVTT
00:00.000 --> 00:01.240
Lâchez, c'est bon.
00:01.240 --> 00:02.240
Ça va?
00:02.240 --> 00:03.240
Oui.
00:03.240 --> 00:04.240
Merci beaucoup.
00:04.240 --> 00:05.240
Chèque ou espèce?
00:05.240 --> 00:08.640
J'ai un chèque sur la commode, il est signé.
00:08.640 --> 00:09.640
Je vais le repirer.
00:09.640 --> 00:10.640
Ok.
00:10.640 --> 00:11.640
Ouh là!
00:11.640 --> 00:12.640
Venez.
00:12.640 --> 00:13.640
Merci.
00:13.640 --> 00:14.640
Ah! C'est qui?
00:14.640 --> 00:21.640
C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
00:21.640 --> 00:26.640
Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
00:26.640 --> 00:27.640
Ça va?
00:27.640 --> 00:44.200
Attendez, tout à l'heure là, c'était vous? Vous? Pas lui? Vous?
00:44.200 --> 00:48.360
Vous avez tout à fait raison, M. Xanaquis, Malek est à l'interne brillant qui apprend
00:48.360 --> 00:49.360
le métier avec moi.
00:49.360 --> 00:50.360
Ah!
00:50.360 --> 00:51.360
Bien.
00:51.360 --> 00:55.520
Justement, il y a la famille Boboune qui m'attend pour une consultation.
00:55.520 --> 00:56.520
Qui?
00:56.520 --> 00:57.760
Faisons pas attendre les Boboune, allez.

BIN
examples/whisperx/sample01.mp4
View File

Binary file not shown.

290
examples/whisperx/sample01.wav.ass
View File

@ -1,290 +0,0 @@
[Script Info]
ScriptType: v4.00+
PlayResX: 384
PlayResY: 288
ScaledBorderAndShadow: yes
[V4+ Styles]
Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
Style: Default,Arial,24,&Hffffff,&Hffffff,&H0,&H0,0,0,0,0,100,100,0,0,1,1,0,2,10,10,10,0
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:1.18,0:00:1.67,Default,,0,0,0,,{\1c&HFF00&\u1}Bella,{\r} Gloria, love.
Dialogue: 0,0:00:1.67,0:00:2.65,Default,,0,0,0,,Bella, Gloria, love.
Dialogue: 0,0:00:2.65,0:00:3.05,Default,,0,0,0,,Bella, {\1c&HFF00&\u1}Gloria,{\r} love.
Dialogue: 0,0:00:3.05,0:00:3.07,Default,,0,0,0,,Bella, Gloria, love.
Dialogue: 0,0:00:3.07,0:00:3.27,Default,,0,0,0,,Bella, Gloria, {\1c&HFF00&\u1}love.{\r}
Dialogue: 0,0:00:3.75,0:00:3.85,Default,,0,0,0,,{\1c&HFF00&\u1}Oh.{\r}
Dialogue: 0,0:00:4.50,0:00:4.72,Default,,0,0,0,,{\1c&HFF00&\u1}How{\r} are you?
Dialogue: 0,0:00:4.72,0:00:5.78,Default,,0,0,0,,How are you?
Dialogue: 0,0:00:5.78,0:00:5.90,Default,,0,0,0,,How {\1c&HFF00&\u1}are{\r} you?
Dialogue: 0,0:00:5.90,0:00:5.94,Default,,0,0,0,,How are you?
Dialogue: 0,0:00:5.94,0:00:6.22,Default,,0,0,0,,How are {\1c&HFF00&\u1}you?{\r}
Dialogue: 0,0:00:6.72,0:00:6.80,Default,,0,0,0,,{\1c&HFF00&\u1}Oh,{\r} I'm OK.
Dialogue: 0,0:00:6.80,0:00:6.88,Default,,0,0,0,,Oh, I'm OK.
Dialogue: 0,0:00:6.88,0:00:7.04,Default,,0,0,0,,Oh, {\1c&HFF00&\u1}I'm{\r} OK.
Dialogue: 0,0:00:7.04,0:00:7.09,Default,,0,0,0,,Oh, I'm OK.
Dialogue: 0,0:00:7.09,0:00:7.13,Default,,0,0,0,,Oh, I'm {\1c&HFF00&\u1}OK.{\r}
Dialogue: 0,0:00:8.41,0:00:8.45,Default,,0,0,0,,{\1c&HFF00&\u1}I{\r} will be.
Dialogue: 0,0:00:8.45,0:00:8.49,Default,,0,0,0,,I will be.
Dialogue: 0,0:00:8.49,0:00:8.73,Default,,0,0,0,,I {\1c&HFF00&\u1}will{\r} be.
Dialogue: 0,0:00:8.73,0:00:8.77,Default,,0,0,0,,I will be.
Dialogue: 0,0:00:8.77,0:00:8.91,Default,,0,0,0,,I will {\1c&HFF00&\u1}be.{\r}
Dialogue: 0,0:00:9.22,0:00:9.30,Default,,0,0,0,,{\1c&HFF00&\u1}I{\r} said she could stay with us tomorrow
Dialogue: 0,0:00:9.30,0:00:9.34,Default,,0,0,0,,I said she could stay with us tomorrow
Dialogue: 0,0:00:9.34,0:00:9.48,Default,,0,0,0,,I {\1c&HFF00&\u1}said{\r} she could stay with us tomorrow
Dialogue: 0,0:00:9.48,0:00:9.52,Default,,0,0,0,,I said she could stay with us tomorrow
Dialogue: 0,0:00:9.52,0:00:9.60,Default,,0,0,0,,I said {\1c&HFF00&\u1}she{\r} could stay with us tomorrow
Dialogue: 0,0:00:9.60,0:00:9.62,Default,,0,0,0,,I said she could stay with us tomorrow
Dialogue: 0,0:00:9.62,0:00:9.76,Default,,0,0,0,,I said she {\1c&HFF00&\u1}could{\r} stay with us tomorrow
Dialogue: 0,0:00:9.76,0:00:9.78,Default,,0,0,0,,I said she could stay with us tomorrow
Dialogue: 0,0:00:9.78,0:00:9.94,Default,,0,0,0,,I said she could {\1c&HFF00&\u1}stay{\r} with us tomorrow
Dialogue: 0,0:00:9.94,0:00:9.96,Default,,0,0,0,,I said she could stay with us tomorrow
Dialogue: 0,0:00:9.96,0:00:10.08,Default,,0,0,0,,I said she could stay {\1c&HFF00&\u1}with{\r} us tomorrow
Dialogue: 0,0:00:10.08,0:00:10.10,Default,,0,0,0,,I said she could stay with us tomorrow
Dialogue: 0,0:00:10.10,0:00:10.14,Default,,0,0,0,,I said she could stay with {\1c&HFF00&\u1}us{\r} tomorrow
Dialogue: 0,0:00:10.14,0:00:10.16,Default,,0,0,0,,I said she could stay with us tomorrow
Dialogue: 0,0:00:10.16,0:00:10.44,Default,,0,0,0,,I said she could stay with us {\1c&HFF00&\u1}tomorrow{\r}
Dialogue: 0,0:00:10.46,0:00:10.54,Default,,0,0,0,,{\1c&HFF00&\u1}just{\r} until she feels better.
Dialogue: 0,0:00:10.54,0:00:10.56,Default,,0,0,0,,just until she feels better.
Dialogue: 0,0:00:10.56,0:00:10.68,Default,,0,0,0,,just {\1c&HFF00&\u1}until{\r} she feels better.
Dialogue: 0,0:00:10.68,0:00:10.70,Default,,0,0,0,,just until she feels better.
Dialogue: 0,0:00:10.70,0:00:10.80,Default,,0,0,0,,just until {\1c&HFF00&\u1}she{\r} feels better.
Dialogue: 0,0:00:10.80,0:00:10.82,Default,,0,0,0,,just until she feels better.
Dialogue: 0,0:00:10.82,0:00:11.05,Default,,0,0,0,,just until she {\1c&HFF00&\u1}feels{\r} better.
Dialogue: 0,0:00:11.05,0:00:11.09,Default,,0,0,0,,just until she feels better.
Dialogue: 0,0:00:11.09,0:00:11.35,Default,,0,0,0,,just until she feels {\1c&HFF00&\u1}better.{\r}
Dialogue: 0,0:00:11.73,0:00:11.95,Default,,0,0,0,,{\1c&HFF00&\u1}Yeah.{\r}
Dialogue: 0,0:00:12.09,0:00:12.17,Default,,0,0,0,,{\1c&HFF00&\u1}Of{\r} course she can.
Dialogue: 0,0:00:12.17,0:00:12.20,Default,,0,0,0,,Of course she can.
Dialogue: 0,0:00:12.20,0:00:12.32,Default,,0,0,0,,Of {\1c&HFF00&\u1}course{\r} she can.
Dialogue: 0,0:00:12.32,0:00:12.38,Default,,0,0,0,,Of course she can.
Dialogue: 0,0:00:12.38,0:00:12.64,Default,,0,0,0,,Of course {\1c&HFF00&\u1}she{\r} can.
Dialogue: 0,0:00:12.64,0:00:12.72,Default,,0,0,0,,Of course she can.
Dialogue: 0,0:00:12.72,0:00:13.24,Default,,0,0,0,,Of course she {\1c&HFF00&\u1}can.{\r}
Dialogue: 0,0:00:13.36,0:00:13.70,Default,,0,0,0,,{\1c&HFF00&\u1}No,{\r} things won't be for long.
Dialogue: 0,0:00:13.70,0:00:13.82,Default,,0,0,0,,No, things won't be for long.
Dialogue: 0,0:00:13.82,0:00:14.12,Default,,0,0,0,,No, {\1c&HFF00&\u1}things{\r} won't be for long.
Dialogue: 0,0:00:14.12,0:00:14.19,Default,,0,0,0,,No, things won't be for long.
Dialogue: 0,0:00:14.19,0:00:14.39,Default,,0,0,0,,No, things {\1c&HFF00&\u1}won't{\r} be for long.
Dialogue: 0,0:00:14.39,0:00:14.43,Default,,0,0,0,,No, things won't be for long.
Dialogue: 0,0:00:14.43,0:00:14.53,Default,,0,0,0,,No, things won't {\1c&HFF00&\u1}be{\r} for long.
Dialogue: 0,0:00:14.53,0:00:14.59,Default,,0,0,0,,No, things won't be for long.
Dialogue: 0,0:00:14.59,0:00:14.73,Default,,0,0,0,,No, things won't be {\1c&HFF00&\u1}for{\r} long.
Dialogue: 0,0:00:14.73,0:00:14.81,Default,,0,0,0,,No, things won't be for long.
Dialogue: 0,0:00:14.81,0:00:15.01,Default,,0,0,0,,No, things won't be for {\1c&HFF00&\u1}long.{\r}
Dialogue: 0,0:00:15.17,0:00:15.41,Default,,0,0,0,,{\1c&HFF00&\u1}Well,{\r} you can stay as long as you want, my love.
Dialogue: 0,0:00:15.41,0:00:15.43,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:15.43,0:00:15.51,Default,,0,0,0,,Well, {\1c&HFF00&\u1}you{\r} can stay as long as you want, my love.
Dialogue: 0,0:00:15.51,0:00:15.55,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:15.55,0:00:15.69,Default,,0,0,0,,Well, you {\1c&HFF00&\u1}can{\r} stay as long as you want, my love.
Dialogue: 0,0:00:15.69,0:00:15.75,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:15.75,0:00:15.89,Default,,0,0,0,,Well, you can {\1c&HFF00&\u1}stay{\r} as long as you want, my love.
Dialogue: 0,0:00:15.89,0:00:15.95,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:15.95,0:00:16.01,Default,,0,0,0,,Well, you can stay {\1c&HFF00&\u1}as{\r} long as you want, my love.
Dialogue: 0,0:00:16.01,0:00:16.05,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:16.05,0:00:16.20,Default,,0,0,0,,Well, you can stay as {\1c&HFF00&\u1}long{\r} as you want, my love.
Dialogue: 0,0:00:16.20,0:00:16.24,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:16.24,0:00:16.28,Default,,0,0,0,,Well, you can stay as long {\1c&HFF00&\u1}as{\r} you want, my love.
Dialogue: 0,0:00:16.28,0:00:16.30,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:16.30,0:00:16.42,Default,,0,0,0,,Well, you can stay as long as {\1c&HFF00&\u1}you{\r} want, my love.
Dialogue: 0,0:00:16.42,0:00:16.46,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:16.46,0:00:16.66,Default,,0,0,0,,Well, you can stay as long as you {\1c&HFF00&\u1}want,{\r} my love.
Dialogue: 0,0:00:16.66,0:00:16.72,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:16.72,0:00:16.92,Default,,0,0,0,,Well, you can stay as long as you want, {\1c&HFF00&\u1}my{\r} love.
Dialogue: 0,0:00:16.92,0:00:16.96,Default,,0,0,0,,Well, you can stay as long as you want, my love.
Dialogue: 0,0:00:16.96,0:00:17.34,Default,,0,0,0,,Well, you can stay as long as you want, my {\1c&HFF00&\u1}love.{\r}
Dialogue: 0,0:00:17.62,0:00:17.86,Default,,0,0,0,,{\1c&HFF00&\u1}I've{\r} really missed you.
Dialogue: 0,0:00:17.86,0:00:17.88,Default,,0,0,0,,I've really missed you.
Dialogue: 0,0:00:17.88,0:00:18.14,Default,,0,0,0,,I've {\1c&HFF00&\u1}really{\r} missed you.
Dialogue: 0,0:00:18.14,0:00:18.19,Default,,0,0,0,,I've really missed you.
Dialogue: 0,0:00:18.19,0:00:18.59,Default,,0,0,0,,I've really {\1c&HFF00&\u1}missed{\r} you.
Dialogue: 0,0:00:18.59,0:00:18.63,Default,,0,0,0,,I've really missed you.
Dialogue: 0,0:00:18.63,0:00:18.81,Default,,0,0,0,,I've really missed {\1c&HFF00&\u1}you.{\r}
Dialogue: 0,0:00:19.49,0:00:19.79,Default,,0,0,0,,{\1c&HFF00&\u1}Pops.{\r}
Dialogue: 0,0:00:20.40,0:00:20.64,Default,,0,0,0,,{\1c&HFF00&\u1}Great{\r} to see you, love.
Dialogue: 0,0:00:20.64,0:00:20.66,Default,,0,0,0,,Great to see you, love.
Dialogue: 0,0:00:20.66,0:00:20.78,Default,,0,0,0,,Great {\1c&HFF00&\u1}to{\r} see you, love.
Dialogue: 0,0:00:20.78,0:00:20.82,Default,,0,0,0,,Great to see you, love.
Dialogue: 0,0:00:20.82,0:00:21.14,Default,,0,0,0,,Great to {\1c&HFF00&\u1}see{\r} you, love.
Dialogue: 0,0:00:21.14,0:00:21.16,Default,,0,0,0,,Great to see you, love.
Dialogue: 0,0:00:21.16,0:00:21.28,Default,,0,0,0,,Great to see {\1c&HFF00&\u1}you,{\r} love.
Dialogue: 0,0:00:21.28,0:00:21.32,Default,,0,0,0,,Great to see you, love.
Dialogue: 0,0:00:21.32,0:00:21.68,Default,,0,0,0,,Great to see you, {\1c&HFF00&\u1}love.{\r}
Dialogue: 0,0:00:21.90,0:00:23.21,Default,,0,0,0,,{\1c&HFF00&\u1}Oh.{\r}
Dialogue: 0,0:00:23.23,0:00:23.29,Default,,0,0,0,,{\1c&HFF00&\u1}All{\r} right, shall we get you off to bed then?
Dialogue: 0,0:00:23.29,0:00:23.31,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:23.31,0:00:23.41,Default,,0,0,0,,All {\1c&HFF00&\u1}right,{\r} shall we get you off to bed then?
Dialogue: 0,0:00:23.41,0:00:23.43,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:23.43,0:00:23.55,Default,,0,0,0,,All right, {\1c&HFF00&\u1}shall{\r} we get you off to bed then?
Dialogue: 0,0:00:23.55,0:00:23.57,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:23.57,0:00:23.65,Default,,0,0,0,,All right, shall {\1c&HFF00&\u1}we{\r} get you off to bed then?
Dialogue: 0,0:00:23.65,0:00:23.67,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:23.67,0:00:23.74,Default,,0,0,0,,All right, shall we {\1c&HFF00&\u1}get{\r} you off to bed then?
Dialogue: 0,0:00:23.74,0:00:23.76,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:23.76,0:00:23.82,Default,,0,0,0,,All right, shall we get {\1c&HFF00&\u1}you{\r} off to bed then?
Dialogue: 0,0:00:23.82,0:00:23.84,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:23.84,0:00:23.94,Default,,0,0,0,,All right, shall we get you {\1c&HFF00&\u1}off{\r} to bed then?
Dialogue: 0,0:00:23.94,0:00:23.96,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:23.96,0:00:24.04,Default,,0,0,0,,All right, shall we get you off {\1c&HFF00&\u1}to{\r} bed then?
Dialogue: 0,0:00:24.04,0:00:24.06,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:24.06,0:00:24.22,Default,,0,0,0,,All right, shall we get you off to {\1c&HFF00&\u1}bed{\r} then?
Dialogue: 0,0:00:24.22,0:00:24.24,Default,,0,0,0,,All right, shall we get you off to bed then?
Dialogue: 0,0:00:24.24,0:00:24.38,Default,,0,0,0,,All right, shall we get you off to bed {\1c&HFF00&\u1}then?{\r}
Dialogue: 0,0:00:24.58,0:00:24.72,Default,,0,0,0,,{\1c&HFF00&\u1}You{\r} should have given me some warm.
Dialogue: 0,0:00:24.72,0:00:24.78,Default,,0,0,0,,You should have given me some warm.
Dialogue: 0,0:00:24.78,0:00:24.98,Default,,0,0,0,,You {\1c&HFF00&\u1}should{\r} have given me some warm.
Dialogue: 0,0:00:24.98,0:00:25.02,Default,,0,0,0,,You should have given me some warm.
Dialogue: 0,0:00:25.02,0:00:25.12,Default,,0,0,0,,You should {\1c&HFF00&\u1}have{\r} given me some warm.
Dialogue: 0,0:00:25.12,0:00:25.16,Default,,0,0,0,,You should have given me some warm.
Dialogue: 0,0:00:25.16,0:00:25.29,Default,,0,0,0,,You should have {\1c&HFF00&\u1}given{\r} me some warm.
Dialogue: 0,0:00:25.29,0:00:25.35,Default,,0,0,0,,You should have given me some warm.
Dialogue: 0,0:00:25.35,0:00:25.45,Default,,0,0,0,,You should have given {\1c&HFF00&\u1}me{\r} some warm.
Dialogue: 0,0:00:25.45,0:00:25.49,Default,,0,0,0,,You should have given me some warm.
Dialogue: 0,0:00:25.49,0:00:25.67,Default,,0,0,0,,You should have given me {\1c&HFF00&\u1}some{\r} warm.
Dialogue: 0,0:00:25.67,0:00:25.81,Default,,0,0,0,,You should have given me some warm.
Dialogue: 0,0:00:25.81,0:00:26.05,Default,,0,0,0,,You should have given me some {\1c&HFF00&\u1}warm.{\r}
Dialogue: 0,0:00:26.31,0:00:26.37,Default,,0,0,0,,{\1c&HFF00&\u1}I{\r} know.
Dialogue: 0,0:00:26.37,0:00:26.39,Default,,0,0,0,,I know.
Dialogue: 0,0:00:26.39,0:00:26.49,Default,,0,0,0,,I {\1c&HFF00&\u1}know.{\r}
Dialogue: 0,0:00:26.61,0:00:26.69,Default,,0,0,0,,{\1c&HFF00&\u1}I'll{\r} have to put the electric blanket on.
Dialogue: 0,0:00:26.69,0:00:26.71,Default,,0,0,0,,I'll have to put the electric blanket on.
Dialogue: 0,0:00:26.71,0:00:26.81,Default,,0,0,0,,I'll {\1c&HFF00&\u1}have{\r} to put the electric blanket on.
Dialogue: 0,0:00:26.81,0:00:26.83,Default,,0,0,0,,I'll have to put the electric blanket on.
Dialogue: 0,0:00:26.83,0:00:27.02,Default,,0,0,0,,I'll have {\1c&HFF00&\u1}to{\r} put the electric blanket on.
Dialogue: 0,0:00:27.02,0:00:27.06,Default,,0,0,0,,I'll have to put the electric blanket on.
Dialogue: 0,0:00:27.06,0:00:27.42,Default,,0,0,0,,I'll have to {\1c&HFF00&\u1}put{\r} the electric blanket on.
Dialogue: 0,0:00:27.42,0:00:27.48,Default,,0,0,0,,I'll have to put the electric blanket on.
Dialogue: 0,0:00:27.48,0:00:27.56,Default,,0,0,0,,I'll have to put {\1c&HFF00&\u1}the{\r} electric blanket on.
Dialogue: 0,0:00:27.56,0:00:27.70,Default,,0,0,0,,I'll have to put the electric blanket on.
Dialogue: 0,0:00:27.70,0:00:28.08,Default,,0,0,0,,I'll have to put the {\1c&HFF00&\u1}electric{\r} blanket on.
Dialogue: 0,0:00:28.08,0:00:28.62,Default,,0,0,0,,I'll have to put the electric blanket on.
Dialogue: 0,0:00:28.62,0:00:28.84,Default,,0,0,0,,I'll have to put the electric {\1c&HFF00&\u1}blanket{\r} on.
Dialogue: 0,0:00:28.84,0:00:28.90,Default,,0,0,0,,I'll have to put the electric blanket on.
Dialogue: 0,0:00:28.90,0:00:28.94,Default,,0,0,0,,I'll have to put the electric blanket {\1c&HFF00&\u1}on.{\r}
Dialogue: 0,0:00:29.49,0:00:29.55,Default,,0,0,0,,{\1c&HFF00&\u1}I'm{\r} sorry.
Dialogue: 0,0:00:29.55,0:00:29.57,Default,,0,0,0,,I'm sorry.
Dialogue: 0,0:00:29.57,0:00:29.82,Default,,0,0,0,,I'm {\1c&HFF00&\u1}sorry.{\r}
Dialogue: 0,0:00:29.98,0:00:30.08,Default,,0,0,0,,{\1c&HFF00&\u1}All{\r} right, Bella.
Dialogue: 0,0:00:30.08,0:00:30.10,Default,,0,0,0,,All right, Bella.
Dialogue: 0,0:00:30.10,0:00:30.29,Default,,0,0,0,,All {\1c&HFF00&\u1}right,{\r} Bella.
Dialogue: 0,0:00:30.29,0:00:30.43,Default,,0,0,0,,All right, Bella.
Dialogue: 0,0:00:30.43,0:00:30.63,Default,,0,0,0,,All right, {\1c&HFF00&\u1}Bella.{\r}
Dialogue: 0,0:00:31.37,0:00:31.58,Default,,0,0,0,,{\1c&HFF00&\u1}Freezing{\r} up there.
Dialogue: 0,0:00:31.58,0:00:31.62,Default,,0,0,0,,Freezing up there.
Dialogue: 0,0:00:31.62,0:00:31.66,Default,,0,0,0,,Freezing {\1c&HFF00&\u1}up{\r} there.
Dialogue: 0,0:00:31.66,0:00:31.68,Default,,0,0,0,,Freezing up there.
Dialogue: 0,0:00:31.68,0:00:31.90,Default,,0,0,0,,Freezing up {\1c&HFF00&\u1}there.{\r}
Dialogue: 0,0:00:31.90,0:00:31.94,Default,,0,0,0,,{\1c&HFF00&\u1}In{\r} a bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:31.94,0:00:31.96,Default,,0,0,0,,In a bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:31.96,0:00:31.98,Default,,0,0,0,,In {\1c&HFF00&\u1}a{\r} bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:31.98,0:00:32.00,Default,,0,0,0,,In a bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:32.00,0:00:32.14,Default,,0,0,0,,In a {\1c&HFF00&\u1}bedroom,{\r} Peter unpacks her suitcase.
Dialogue: 0,0:00:32.14,0:00:32.20,Default,,0,0,0,,In a bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:32.20,0:00:32.50,Default,,0,0,0,,In a bedroom, {\1c&HFF00&\u1}Peter{\r} unpacks her suitcase.
Dialogue: 0,0:00:32.50,0:00:32.58,Default,,0,0,0,,In a bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:32.58,0:00:32.98,Default,,0,0,0,,In a bedroom, Peter {\1c&HFF00&\u1}unpacks{\r} her suitcase.
Dialogue: 0,0:00:32.98,0:00:33.00,Default,,0,0,0,,In a bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:33.00,0:00:33.10,Default,,0,0,0,,In a bedroom, Peter unpacks {\1c&HFF00&\u1}her{\r} suitcase.
Dialogue: 0,0:00:33.10,0:00:33.16,Default,,0,0,0,,In a bedroom, Peter unpacks her suitcase.
Dialogue: 0,0:00:33.16,0:00:33.65,Default,,0,0,0,,In a bedroom, Peter unpacks her {\1c&HFF00&\u1}suitcase.{\r}
Dialogue: 0,0:00:34.27,0:00:34.35,Default,,0,0,0,,{\1c&HFF00&\u1}The{\r} middle-aged woman opens her green case.
Dialogue: 0,0:00:34.35,0:00:34.39,Default,,0,0,0,,The middle-aged woman opens her green case.
Dialogue: 0,0:00:34.39,0:00:34.91,Default,,0,0,0,,The {\1c&HFF00&\u1}middle-aged{\r} woman opens her green case.
Dialogue: 0,0:00:34.91,0:00:34.99,Default,,0,0,0,,The middle-aged woman opens her green case.
Dialogue: 0,0:00:34.99,0:00:35.27,Default,,0,0,0,,The middle-aged {\1c&HFF00&\u1}woman{\r} opens her green case.
Dialogue: 0,0:00:35.27,0:00:35.39,Default,,0,0,0,,The middle-aged woman opens her green case.
Dialogue: 0,0:00:35.39,0:00:35.67,Default,,0,0,0,,The middle-aged woman {\1c&HFF00&\u1}opens{\r} her green case.
Dialogue: 0,0:00:35.67,0:00:35.71,Default,,0,0,0,,The middle-aged woman opens her green case.
Dialogue: 0,0:00:35.71,0:00:35.81,Default,,0,0,0,,The middle-aged woman opens {\1c&HFF00&\u1}her{\r} green case.
Dialogue: 0,0:00:35.81,0:00:35.85,Default,,0,0,0,,The middle-aged woman opens her green case.
Dialogue: 0,0:00:35.85,0:00:36.19,Default,,0,0,0,,The middle-aged woman opens her {\1c&HFF00&\u1}green{\r} case.
Dialogue: 0,0:00:36.19,0:00:36.23,Default,,0,0,0,,The middle-aged woman opens her green case.
Dialogue: 0,0:00:36.23,0:00:36.53,Default,,0,0,0,,The middle-aged woman opens her green {\1c&HFF00&\u1}case.{\r}
Dialogue: 0,0:00:38.13,0:00:38.25,Default,,0,0,0,,{\1c&HFF00&\u1}Do{\r} you want your PJs?
Dialogue: 0,0:00:38.25,0:00:38.28,Default,,0,0,0,,Do you want your PJs?
Dialogue: 0,0:00:38.28,0:00:38.36,Default,,0,0,0,,Do {\1c&HFF00&\u1}you{\r} want your PJs?
Dialogue: 0,0:00:38.36,0:00:38.38,Default,,0,0,0,,Do you want your PJs?
Dialogue: 0,0:00:38.38,0:00:38.54,Default,,0,0,0,,Do you {\1c&HFF00&\u1}want{\r} your PJs?
Dialogue: 0,0:00:38.54,0:00:38.56,Default,,0,0,0,,Do you want your PJs?
Dialogue: 0,0:00:38.56,0:00:38.74,Default,,0,0,0,,Do you want {\1c&HFF00&\u1}your{\r} PJs?
Dialogue: 0,0:00:38.74,0:00:38.88,Default,,0,0,0,,Do you want your PJs?
Dialogue: 0,0:00:38.88,0:00:39.30,Default,,0,0,0,,Do you want your {\1c&HFF00&\u1}PJs?{\r}
Dialogue: 0,0:00:39.88,0:00:40.18,Default,,0,0,0,,{\1c&HFF00&\u1}Yeah.{\r}
Dialogue: 0,0:00:42.39,0:00:42.69,Default,,0,0,0,,{\1c&HFF00&\u1}Lifting{\r} a bundle of pajamas,
Dialogue: 0,0:00:42.69,0:00:42.73,Default,,0,0,0,,Lifting a bundle of pajamas,
Dialogue: 0,0:00:42.73,0:00:42.75,Default,,0,0,0,,Lifting {\1c&HFF00&\u1}a{\r} bundle of pajamas,
Dialogue: 0,0:00:42.75,0:00:42.81,Default,,0,0,0,,Lifting a bundle of pajamas,
Dialogue: 0,0:00:42.81,0:00:43.11,Default,,0,0,0,,Lifting a {\1c&HFF00&\u1}bundle{\r} of pajamas,
Dialogue: 0,0:00:43.11,0:00:43.13,Default,,0,0,0,,Lifting a bundle of pajamas,
Dialogue: 0,0:00:43.13,0:00:43.19,Default,,0,0,0,,Lifting a bundle {\1c&HFF00&\u1}of{\r} pajamas,
Dialogue: 0,0:00:43.19,0:00:43.25,Default,,0,0,0,,Lifting a bundle of pajamas,
Dialogue: 0,0:00:43.25,0:00:43.77,Default,,0,0,0,,Lifting a bundle of {\1c&HFF00&\u1}pajamas,{\r}
Dialogue: 0,0:00:44.07,0:00:44.31,Default,,0,0,0,,{\1c&HFF00&\u1}Peter{\r} finds a sheet of paper labeled
Dialogue: 0,0:00:44.31,0:00:44.37,Default,,0,0,0,,Peter finds a sheet of paper labeled
Dialogue: 0,0:00:44.37,0:00:44.63,Default,,0,0,0,,Peter {\1c&HFF00&\u1}finds{\r} a sheet of paper labeled
Dialogue: 0,0:00:44.63,0:00:44.67,Default,,0,0,0,,Peter finds a sheet of paper labeled
Dialogue: 0,0:00:44.67,0:00:44.69,Default,,0,0,0,,Peter finds {\1c&HFF00&\u1}a{\r} sheet of paper labeled
Dialogue: 0,0:00:44.69,0:00:44.75,Default,,0,0,0,,Peter finds a sheet of paper labeled
Dialogue: 0,0:00:44.75,0:00:44.95,Default,,0,0,0,,Peter finds a {\1c&HFF00&\u1}sheet{\r} of paper labeled
Dialogue: 0,0:00:44.95,0:00:44.99,Default,,0,0,0,,Peter finds a sheet of paper labeled
Dialogue: 0,0:00:44.99,0:00:45.05,Default,,0,0,0,,Peter finds a sheet {\1c&HFF00&\u1}of{\r} paper labeled
Dialogue: 0,0:00:45.05,0:00:45.11,Default,,0,0,0,,Peter finds a sheet of paper labeled
Dialogue: 0,0:00:45.11,0:00:45.46,Default,,0,0,0,,Peter finds a sheet of {\1c&HFF00&\u1}paper{\r} labeled
Dialogue: 0,0:00:45.46,0:00:45.54,Default,,0,0,0,,Peter finds a sheet of paper labeled
Dialogue: 0,0:00:45.54,0:00:45.88,Default,,0,0,0,,Peter finds a sheet of paper {\1c&HFF00&\u1}labeled{\r}
Dialogue: 0,0:00:46.34,0:00:47.04,Default,,0,0,0,,{\1c&HFF00&\u1}Lancaster{\r} North Hospital discharge sheet.
Dialogue: 0,0:00:47.04,0:00:47.12,Default,,0,0,0,,Lancaster North Hospital discharge sheet.
Dialogue: 0,0:00:47.12,0:00:47.38,Default,,0,0,0,,Lancaster {\1c&HFF00&\u1}North{\r} Hospital discharge sheet.
Dialogue: 0,0:00:47.38,0:00:47.44,Default,,0,0,0,,Lancaster North Hospital discharge sheet.
Dialogue: 0,0:00:47.44,0:00:47.94,Default,,0,0,0,,Lancaster North {\1c&HFF00&\u1}Hospital{\r} discharge sheet.
Dialogue: 0,0:00:47.94,0:00:48.27,Default,,0,0,0,,Lancaster North Hospital discharge sheet.
Dialogue: 0,0:00:48.27,0:00:48.93,Default,,0,0,0,,Lancaster North Hospital {\1c&HFF00&\u1}discharge{\r} sheet.
Dialogue: 0,0:00:48.93,0:00:49.03,Default,,0,0,0,,Lancaster North Hospital discharge sheet.
Dialogue: 0,0:00:49.03,0:00:49.25,Default,,0,0,0,,Lancaster North Hospital discharge {\1c&HFF00&\u1}sheet.{\r}
Dialogue: 0,0:00:50.29,0:00:50.37,Default,,0,0,0,,{\1c&HFF00&\u1}He{\r} closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:50.37,0:00:50.41,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:50.41,0:00:50.77,Default,,0,0,0,,He {\1c&HFF00&\u1}closes{\r} the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:50.77,0:00:50.81,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:50.81,0:00:50.91,Default,,0,0,0,,He closes {\1c&HFF00&\u1}the{\r} suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:50.91,0:00:50.95,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:50.95,0:00:51.39,Default,,0,0,0,,He closes the {\1c&HFF00&\u1}suitcase{\r} and brings Gloria the pajamas.
Dialogue: 0,0:00:51.39,0:00:51.43,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:51.43,0:00:51.51,Default,,0,0,0,,He closes the suitcase {\1c&HFF00&\u1}and{\r} brings Gloria the pajamas.
Dialogue: 0,0:00:51.51,0:00:51.53,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:51.53,0:00:51.79,Default,,0,0,0,,He closes the suitcase and {\1c&HFF00&\u1}brings{\r} Gloria the pajamas.
Dialogue: 0,0:00:51.79,0:00:51.83,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:51.83,0:00:52.23,Default,,0,0,0,,He closes the suitcase and brings {\1c&HFF00&\u1}Gloria{\r} the pajamas.
Dialogue: 0,0:00:52.23,0:00:52.25,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:52.25,0:00:52.32,Default,,0,0,0,,He closes the suitcase and brings Gloria {\1c&HFF00&\u1}the{\r} pajamas.
Dialogue: 0,0:00:52.32,0:00:52.36,Default,,0,0,0,,He closes the suitcase and brings Gloria the pajamas.
Dialogue: 0,0:00:52.36,0:00:52.86,Default,,0,0,0,,He closes the suitcase and brings Gloria the {\1c&HFF00&\u1}pajamas.{\r}
Dialogue: 0,0:00:54.19,0:00:54.49,Default,,0,0,0,,{\1c&HFF00&\u1}There{\r} you go.
Dialogue: 0,0:00:54.49,0:00:54.55,Default,,0,0,0,,There you go.
Dialogue: 0,0:00:54.55,0:00:54.77,Default,,0,0,0,,There {\1c&HFF00&\u1}you{\r} go.
Dialogue: 0,0:00:54.77,0:00:54.79,Default,,0,0,0,,There you go.
Dialogue: 0,0:00:54.79,0:00:54.83,Default,,0,0,0,,There you {\1c&HFF00&\u1}go.{\r}
Dialogue: 0,0:00:55.65,0:00:55.77,Default,,0,0,0,,{\1c&HFF00&\u1}Thank{\r} you.
Dialogue: 0,0:00:55.77,0:00:55.80,Default,,0,0,0,,Thank you.
Dialogue: 0,0:00:55.80,0:00:55.90,Default,,0,0,0,,Thank {\1c&HFF00&\u1}you.{\r}
Dialogue: 0,0:00:55.90,0:00:55.94,Default,,0,0,0,,{\1c&HFF00&\u1}He{\r} picks up the locket.
Dialogue: 0,0:00:55.94,0:00:55.96,Default,,0,0,0,,He picks up the locket.
Dialogue: 0,0:00:55.96,0:00:56.10,Default,,0,0,0,,He {\1c&HFF00&\u1}picks{\r} up the locket.
Dialogue: 0,0:00:56.10,0:00:56.12,Default,,0,0,0,,He picks up the locket.
Dialogue: 0,0:00:56.12,0:00:56.20,Default,,0,0,0,,He picks {\1c&HFF00&\u1}up{\r} the locket.
Dialogue: 0,0:00:56.20,0:00:56.22,Default,,0,0,0,,He picks up the locket.
Dialogue: 0,0:00:56.22,0:00:56.32,Default,,0,0,0,,He picks up {\1c&HFF00&\u1}the{\r} locket.
Dialogue: 0,0:00:56.32,0:00:56.36,Default,,0,0,0,,He picks up the locket.
Dialogue: 0,0:00:56.36,0:00:56.74,Default,,0,0,0,,He picks up the {\1c&HFF00&\u1}locket.{\r}
Dialogue: 0,0:00:57.12,0:00:57.22,Default,,0,0,0,,{\1c&HFF00&\u1}You{\r} kept it.
Dialogue: 0,0:00:57.22,0:00:57.27,Default,,0,0,0,,You kept it.
Dialogue: 0,0:00:57.27,0:00:57.47,Default,,0,0,0,,You {\1c&HFF00&\u1}kept{\r} it.
Dialogue: 0,0:00:57.47,0:00:57.55,Default,,0,0,0,,You kept it.
Dialogue: 0,0:00:57.55,0:00:57.63,Default,,0,0,0,,You kept {\1c&HFF00&\u1}it.{\r}
Dialogue: 0,0:00:58.87,0:00:58.99,Default,,0,0,0,,{\1c&HFF00&\u1}Oh,{\r} of course.
Dialogue: 0,0:00:58.99,0:00:59.28,Default,,0,0,0,,Oh, of course.
Dialogue: 0,0:00:59.28,0:00:59.58,Default,,0,0,0,,Oh, {\1c&HFF00&\u1}of{\r} course.
Dialogue: 0,0:00:59.58,0:00:59.68,Default,,0,0,0,,Oh, of course.
Dialogue: 0,0:00:59.68,0:00:59.96,Default,,0,0,0,,Oh, of {\1c&HFF00&\u1}course.{\r}

140
examples/whisperx/sample01.wav.srt
View File

@ -1,140 +0,0 @@
1
00:00:01,185 --> 00:00:03,273
Bella, Gloria, love.
2
00:00:03,754 --> 00:00:03,855
Oh.
3
00:00:04,496 --> 00:00:06,219
How are you?
4
00:00:06,723 --> 00:00:07,126
Oh, I'm OK.
5
00:00:08,412 --> 00:00:08,915
I will be.
6
00:00:09,215 --> 00:00:10,439
I said she could stay with us tomorrow
7
00:00:10,459 --> 00:00:11,351
just until she feels better.
8
00:00:11,733 --> 00:00:11,954
Yeah.
9
00:00:12,095 --> 00:00:13,238
Of course she can.
10
00:00:13,359 --> 00:00:15,012
No, things won't be for long.
11
00:00:15,173 --> 00:00:17,338
Well, you can stay as long as you want, my love.
12
00:00:17,621 --> 00:00:18,810
I've really missed you.
13
00:00:19,493 --> 00:00:19,795
Pops.
14
00:00:20,396 --> 00:00:21,679
Great to see you, love.
15
00:00:21,901 --> 00:00:23,213
Oh.
16
00:00:23,233 --> 00:00:24,378
All right, shall we get you off to bed then?
17
00:00:24,579 --> 00:00:26,052
You should have given me some warm.
18
00:00:26,313 --> 00:00:26,494
I know.
19
00:00:26,614 --> 00:00:28,940
I'll have to put the electric blanket on.
20
00:00:29,490 --> 00:00:29,817
I'm sorry.
21
00:00:29,980 --> 00:00:30,633
All right, Bella.
22
00:00:31,375 --> 00:00:31,897
Freezing up there.
23
00:00:31,897 --> 00:00:33,647
In a bedroom, Peter unpacks her suitcase.
24
00:00:34,268 --> 00:00:36,533
The middle-aged woman opens her green case.
25
00:00:38,135 --> 00:00:39,296
Do you want your PJs?
26
00:00:39,879 --> 00:00:40,181
Yeah.
27
00:00:42,388 --> 00:00:43,773
Lifting a bundle of pajamas,
28
00:00:44,073 --> 00:00:45,876
Peter finds a sheet of paper labeled
29
00:00:46,338 --> 00:00:49,249
Lancaster North Hospital discharge sheet.
30
00:00:50,291 --> 00:00:52,856
He closes the suitcase and brings Gloria the pajamas.
31
00:00:54,186 --> 00:00:54,831
There you go.
32
00:00:55,654 --> 00:00:55,895
Thank you.
33
00:00:55,895 --> 00:00:56,742
He picks up the locket.
34
00:00:57,124 --> 00:00:57,627
You kept it.
35
00:00:58,874 --> 00:00:59,960
Oh, of course.

624
examples/whisperx/sample01.wav.word.srt
View File

@ -1,624 +0,0 @@
1
00:00:01,185 --> 00:00:01,667
Bella,
2
00:00:02,651 --> 00:00:03,052
Gloria,
3
00:00:03,072 --> 00:00:03,273
love.
4
00:00:03,754 --> 00:00:03,855
Oh.
5
00:00:04,496 --> 00:00:04,716
How
6
00:00:05,778 --> 00:00:05,898
are
7
00:00:05,938 --> 00:00:06,219
you?
8
00:00:06,723 --> 00:00:06,803
Oh,
9
00:00:06,884 --> 00:00:07,045
I'm
10
00:00:07,085 --> 00:00:07,126
OK.
11
00:00:08,412 --> 00:00:08,452
I
12
00:00:08,492 --> 00:00:08,734
will
13
00:00:08,774 --> 00:00:08,915
be.
14
00:00:09,215 --> 00:00:09,296
I
15
00:00:09,336 --> 00:00:09,476
said
16
00:00:09,516 --> 00:00:09,596
she
17
00:00:09,616 --> 00:00:09,757
could
18
00:00:09,777 --> 00:00:09,937
stay
19
00:00:09,957 --> 00:00:10,078
with
20
00:00:10,098 --> 00:00:10,138
us
21
00:00:10,158 --> 00:00:10,439
tomorrow
22
00:00:10,459 --> 00:00:10,540
just
23
00:00:10,560 --> 00:00:10,682
until
24
00:00:10,702 --> 00:00:10,804
she
25
00:00:10,824 --> 00:00:11,047
feels
26
00:00:11,087 --> 00:00:11,351
better.
27
00:00:11,733 --> 00:00:11,954
Yeah.
28
00:00:12,095 --> 00:00:12,175
Of
29
00:00:12,195 --> 00:00:12,315
course
30
00:00:12,376 --> 00:00:12,636
she
31
00:00:12,716 --> 00:00:13,238
can.
32
00:00:13,359 --> 00:00:13,702
No,
33
00:00:13,823 --> 00:00:14,125
things
34
00:00:14,185 --> 00:00:14,387
won't
35
00:00:14,427 --> 00:00:14,528
be
36
00:00:14,589 --> 00:00:14,730
for
37
00:00:14,810 --> 00:00:15,012
long.
38
00:00:15,173 --> 00:00:15,413
Well,
39
00:00:15,433 --> 00:00:15,513
you
40
00:00:15,554 --> 00:00:15,694
can
41
00:00:15,754 --> 00:00:15,894
stay
42
00:00:15,955 --> 00:00:16,015
as
43
00:00:16,055 --> 00:00:16,195
long
44
00:00:16,235 --> 00:00:16,275
as
45
00:00:16,295 --> 00:00:16,416
you
46
00:00:16,456 --> 00:00:16,656
want,
47
00:00:16,717 --> 00:00:16,917
my
48
00:00:16,957 --> 00:00:17,338
love.
49
00:00:17,621 --> 00:00:17,863
I've
50
00:00:17,883 --> 00:00:18,145
really
51
00:00:18,185 --> 00:00:18,588
missed
52
00:00:18,629 --> 00:00:18,810
you.
53
00:00:19,493 --> 00:00:19,795
Pops.
54
00:00:20,396 --> 00:00:20,637
Great
55
00:00:20,657 --> 00:00:20,777
to
56
00:00:20,817 --> 00:00:21,138
see
57
00:00:21,158 --> 00:00:21,278
you,
58
00:00:21,318 --> 00:00:21,679
love.
59
00:00:21,901 --> 00:00:23,213
Oh.
60
00:00:23,233 --> 00:00:23,293
All
61
00:00:23,313 --> 00:00:23,414
right,
62
00:00:23,434 --> 00:00:23,554
shall
63
00:00:23,574 --> 00:00:23,655
we
64
00:00:23,675 --> 00:00:23,735
get
65
00:00:23,755 --> 00:00:23,815
you
66
00:00:23,835 --> 00:00:23,936
off
67
00:00:23,956 --> 00:00:24,036
to
68
00:00:24,056 --> 00:00:24,217
bed
69
00:00:24,237 --> 00:00:24,378
then?
70
00:00:24,579 --> 00:00:24,720
You
71
00:00:24,781 --> 00:00:24,983
should
72
00:00:25,023 --> 00:00:25,124
have
73
00:00:25,164 --> 00:00:25,285
given
74
00:00:25,346 --> 00:00:25,447
me
75
00:00:25,487 --> 00:00:25,669
some
76
00:00:25,810 --> 00:00:26,052
warm.
77
00:00:26,313 --> 00:00:26,373
I
78
00:00:26,393 --> 00:00:26,494
know.
79
00:00:26,614 --> 00:00:26,694
I'll
80
00:00:26,714 --> 00:00:26,815
have
81
00:00:26,835 --> 00:00:27,015
to
82
00:00:27,055 --> 00:00:27,416
put
83
00:00:27,476 --> 00:00:27,556
the
84
00:00:27,697 --> 00:00:28,078
electric
85
00:00:28,619 --> 00:00:28,840
blanket
86
00:00:28,900 --> 00:00:28,940
on.
87
00:00:29,490 --> 00:00:29,551
I'm
88
00:00:29,572 --> 00:00:29,817
sorry.
89
00:00:29,980 --> 00:00:30,082
All
90
00:00:30,102 --> 00:00:30,286
right,
91
00:00:30,429 --> 00:00:30,633
Bella.
92
00:00:31,375 --> 00:00:31,576
Freezing
93
00:00:31,616 --> 00:00:31,656
up
94
00:00:31,676 --> 00:00:31,897
there.
95
00:00:31,897 --> 00:00:31,937
In
96
00:00:31,957 --> 00:00:31,977
a
97
00:00:31,997 --> 00:00:32,138
bedroom,
98
00:00:32,198 --> 00:00:32,500
Peter
99
00:00:32,581 --> 00:00:32,983
unpacks
100
00:00:33,003 --> 00:00:33,103
her
101
00:00:33,164 --> 00:00:33,647
suitcase.
102
00:00:34,268 --> 00:00:34,348
The
103
00:00:34,388 --> 00:00:34,909
middle-aged
104
00:00:34,989 --> 00:00:35,270
woman
105
00:00:35,390 --> 00:00:35,671
opens
106
00:00:35,711 --> 00:00:35,811
her
107
00:00:35,851 --> 00:00:36,192
green
108
00:00:36,232 --> 00:00:36,533
case.
109
00:00:38,135 --> 00:00:38,255
Do
110
00:00:38,275 --> 00:00:38,355
you
111
00:00:38,375 --> 00:00:38,535
want
112
00:00:38,555 --> 00:00:38,736
your
113
00:00:38,876 --> 00:00:39,296
PJs?
114
00:00:39,879 --> 00:00:40,181
Yeah.
115
00:00:42,388 --> 00:00:42,689
Lifting
116
00:00:42,729 --> 00:00:42,749
a
117
00:00:42,809 --> 00:00:43,110
bundle
118
00:00:43,131 --> 00:00:43,191
of
119
00:00:43,251 --> 00:00:43,773
pajamas,
120
00:00:44,073 --> 00:00:44,314
Peter
121
00:00:44,374 --> 00:00:44,634
finds
122
00:00:44,674 --> 00:00:44,694
a
123
00:00:44,754 --> 00:00:44,955
sheet
124
00:00:44,995 --> 00:00:45,055
of
125
00:00:45,115 --> 00:00:45,456
paper
126
00:00:45,536 --> 00:00:45,876
labeled
127
00:00:46,338 --> 00:00:47,041
Lancaster
128
00:00:47,121 --> 00:00:47,382
North
129
00:00:47,442 --> 00:00:47,944
Hospital
130
00:00:48,266 --> 00:00:48,928
discharge
131
00:00:49,029 --> 00:00:49,249
sheet.
132
00:00:50,291 --> 00:00:50,371
He
133
00:00:50,412 --> 00:00:50,772
closes
134
00:00:50,812 --> 00:00:50,912
the
135
00:00:50,953 --> 00:00:51,393
suitcase
136
00:00:51,433 --> 00:00:51,514
and
137
00:00:51,534 --> 00:00:51,794
brings
138
00:00:51,834 --> 00:00:52,235
Gloria
139
00:00:52,255 --> 00:00:52,315
the
140
00:00:52,355 --> 00:00:52,856
pajamas.
141
00:00:54,186 --> 00:00:54,488
There
142
00:00:54,549 --> 00:00:54,771
you
143
00:00:54,791 --> 00:00:54,831
go.
144
00:00:55,654 --> 00:00:55,775
Thank
145
00:00:55,795 --> 00:00:55,895
you.
146
00:00:55,895 --> 00:00:55,936
He
147
00:00:55,956 --> 00:00:56,097
picks
148
00:00:56,117 --> 00:00:56,198
up
149
00:00:56,218 --> 00:00:56,319
the
150
00:00:56,359 --> 00:00:56,742
locket.
151
00:00:57,124 --> 00:00:57,225
You
152
00:00:57,265 --> 00:00:57,466
kept
153
00:00:57,547 --> 00:00:57,627
it.
154
00:00:58,874 --> 00:00:58,994
Oh,
155
00:00:59,276 --> 00:00:59,578
of
156
00:00:59,678 --> 00:00:59,960
course.

184
examples/whisperx/sample_de_01.wav.ass
View File

@ -1,184 +0,0 @@
[Script Info]
ScriptType: v4.00+
PlayResX: 384
PlayResY: 288
ScaledBorderAndShadow: yes
[V4+ Styles]
Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
Style: Default,Arial,24,&Hffffff,&Hffffff,&H0,&H0,0,0,0,0,100,100,0,0,1,1,0,2,10,10,10,0
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:0.56,0:00:0.96,Default,,0,0,0,,{\1c&HFF00&\u1}Weinlein{\r} von Hammersmann
Dialogue: 0,0:00:0.96,0:00:0.98,Default,,0,0,0,,Weinlein von Hammersmann
Dialogue: 0,0:00:0.98,0:00:1.21,Default,,0,0,0,,Weinlein {\1c&HFF00&\u1}von{\r} Hammersmann
Dialogue: 0,0:00:1.21,0:00:1.23,Default,,0,0,0,,Weinlein von Hammersmann
Dialogue: 0,0:00:1.23,0:00:1.87,Default,,0,0,0,,Weinlein von {\1c&HFF00&\u1}Hammersmann{\r}
Dialogue: 0,0:00:6.19,0:00:6.61,Default,,0,0,0,,{\1c&HFF00&\u1}Oberst{\r} Lande, es ist lange her
Dialogue: 0,0:00:6.61,0:00:6.63,Default,,0,0,0,,Oberst Lande, es ist lange her
Dialogue: 0,0:00:6.63,0:00:6.97,Default,,0,0,0,,Oberst {\1c&HFF00&\u1}Lande,{\r} es ist lange her
Dialogue: 0,0:00:6.97,0:00:7.01,Default,,0,0,0,,Oberst Lande, es ist lange her
Dialogue: 0,0:00:7.01,0:00:7.17,Default,,0,0,0,,Oberst Lande, {\1c&HFF00&\u1}es{\r} ist lange her
Dialogue: 0,0:00:7.17,0:00:7.21,Default,,0,0,0,,Oberst Lande, es ist lange her
Dialogue: 0,0:00:7.21,0:00:7.47,Default,,0,0,0,,Oberst Lande, es {\1c&HFF00&\u1}ist{\r} lange her
Dialogue: 0,0:00:7.47,0:00:7.49,Default,,0,0,0,,Oberst Lande, es ist lange her
Dialogue: 0,0:00:7.49,0:00:7.77,Default,,0,0,0,,Oberst Lande, es ist {\1c&HFF00&\u1}lange{\r} her
Dialogue: 0,0:00:7.77,0:00:7.83,Default,,0,0,0,,Oberst Lande, es ist lange her
Dialogue: 0,0:00:7.83,0:00:8.01,Default,,0,0,0,,Oberst Lande, es ist lange {\1c&HFF00&\u1}her{\r}
Dialogue: 0,0:00:8.01,0:00:8.17,Default,,0,0,0,,{\1c&HFF00&\u1}Schneide{\r} ich wie eh und je
Dialogue: 0,0:00:8.17,0:00:8.19,Default,,0,0,0,,Schneide ich wie eh und je
Dialogue: 0,0:00:8.19,0:00:8.25,Default,,0,0,0,,Schneide {\1c&HFF00&\u1}ich{\r} wie eh und je
Dialogue: 0,0:00:8.25,0:00:8.27,Default,,0,0,0,,Schneide ich wie eh und je
Dialogue: 0,0:00:8.27,0:00:8.33,Default,,0,0,0,,Schneide ich {\1c&HFF00&\u1}wie{\r} eh und je
Dialogue: 0,0:00:8.33,0:00:8.35,Default,,0,0,0,,Schneide ich wie eh und je
Dialogue: 0,0:00:8.35,0:00:8.39,Default,,0,0,0,,Schneide ich wie {\1c&HFF00&\u1}eh{\r} und je
Dialogue: 0,0:00:8.39,0:00:8.41,Default,,0,0,0,,Schneide ich wie eh und je
Dialogue: 0,0:00:8.41,0:00:8.47,Default,,0,0,0,,Schneide ich wie eh {\1c&HFF00&\u1}und{\r} je
Dialogue: 0,0:00:8.47,0:00:8.49,Default,,0,0,0,,Schneide ich wie eh und je
Dialogue: 0,0:00:8.49,0:00:8.53,Default,,0,0,0,,Schneide ich wie eh und {\1c&HFF00&\u1}je{\r}
Dialogue: 0,0:00:13.99,0:00:14.17,Default,,0,0,0,,{\1c&HFF00&\u1}Also{\r} was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.17,0:00:14.21,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.21,0:00:14.45,Default,,0,0,0,,Also {\1c&HFF00&\u1}was{\r} ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.45,0:00:14.47,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.47,0:00:14.59,Default,,0,0,0,,Also was {\1c&HFF00&\u1}ist{\r} mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.59,0:00:14.61,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.61,0:00:14.71,Default,,0,0,0,,Also was ist {\1c&HFF00&\u1}mit{\r} Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.71,0:00:14.75,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.75,0:00:14.91,Default,,0,0,0,,Also was ist mit {\1c&HFF00&\u1}Ihrem{\r} wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.91,0:00:14.93,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:14.93,0:00:15.55,Default,,0,0,0,,Also was ist mit Ihrem {\1c&HFF00&\u1}wunderschönen{\r} Bein geschehen?
Dialogue: 0,0:00:15.55,0:00:15.59,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:15.59,0:00:15.81,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen {\1c&HFF00&\u1}Bein{\r} geschehen?
Dialogue: 0,0:00:15.81,0:00:15.85,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein geschehen?
Dialogue: 0,0:00:15.85,0:00:16.23,Default,,0,0,0,,Also was ist mit Ihrem wunderschönen Bein {\1c&HFF00&\u1}geschehen?{\r}
Dialogue: 0,0:00:17.03,0:00:17.15,Default,,0,0,0,,{\1c&HFF00&\u1}Ein{\r} Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:17.15,0:00:17.17,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:17.17,0:00:17.74,Default,,0,0,0,,Ein {\1c&HFF00&\u1}Nebenprodukt{\r} der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:17.74,0:00:17.78,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:17.78,0:00:17.92,Default,,0,0,0,,Ein Nebenprodukt {\1c&HFF00&\u1}der{\r} Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:17.92,0:00:17.96,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:17.96,0:00:18.60,Default,,0,0,0,,Ein Nebenprodukt der {\1c&HFF00&\u1}Arschtritte,{\r} die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.60,0:00:18.64,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.64,0:00:18.70,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, {\1c&HFF00&\u1}die{\r} Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.70,0:00:18.74,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.74,0:00:18.82,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die {\1c&HFF00&\u1}Sie{\r} in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.82,0:00:18.86,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.86,0:00:18.92,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie {\1c&HFF00&\u1}in{\r} der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.92,0:00:18.94,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:18.94,0:00:19.02,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in {\1c&HFF00&\u1}der{\r} deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:19.02,0:00:19.04,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:19.04,0:00:19.32,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der {\1c&HFF00&\u1}deutschen{\r} Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:19.32,0:00:19.36,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:19.36,0:00:19.88,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen {\1c&HFF00&\u1}Filmwelt{\r} ausleihen, zweifelsohne
Dialogue: 0,0:00:19.88,0:00:19.94,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:19.94,0:00:20.48,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt {\1c&HFF00&\u1}ausleihen,{\r} zweifelsohne
Dialogue: 0,0:00:20.48,0:00:20.52,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
Dialogue: 0,0:00:20.52,0:00:21.22,Default,,0,0,0,,Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, {\1c&HFF00&\u1}zweifelsohne{\r}
Dialogue: 0,0:00:22.10,0:00:22.36,Default,,0,0,0,,{\1c&HFF00&\u1}Sparen{\r} Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.36,0:00:22.38,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.38,0:00:22.48,Default,,0,0,0,,Sparen {\1c&HFF00&\u1}Sie{\r} sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.48,0:00:22.50,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.50,0:00:22.62,Default,,0,0,0,,Sparen Sie {\1c&HFF00&\u1}sich{\r} Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.62,0:00:22.64,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.64,0:00:22.83,Default,,0,0,0,,Sparen Sie sich {\1c&HFF00&\u1}Ihre{\r} Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.83,0:00:22.87,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:22.87,0:00:23.43,Default,,0,0,0,,Sparen Sie sich Ihre {\1c&HFF00&\u1}Komplimente,{\r} Sie alter Hund
Dialogue: 0,0:00:23.43,0:00:23.45,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:23.45,0:00:23.55,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, {\1c&HFF00&\u1}Sie{\r} alter Hund
Dialogue: 0,0:00:23.55,0:00:23.59,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:23.59,0:00:23.89,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie {\1c&HFF00&\u1}alter{\r} Hund
Dialogue: 0,0:00:23.89,0:00:23.93,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter Hund
Dialogue: 0,0:00:23.93,0:00:24.05,Default,,0,0,0,,Sparen Sie sich Ihre Komplimente, Sie alter {\1c&HFF00&\u1}Hund{\r}
Dialogue: 0,0:00:24.57,0:00:24.65,Default,,0,0,0,,{\1c&HFF00&\u1}Ich{\r} kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:24.65,0:00:24.67,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:24.67,0:00:24.97,Default,,0,0,0,,Ich {\1c&HFF00&\u1}kenne{\r} zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:24.97,0:00:25.01,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.01,0:00:25.19,Default,,0,0,0,,Ich kenne {\1c&HFF00&\u1}zu{\r} viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.19,0:00:25.23,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.23,0:00:25.49,Default,,0,0,0,,Ich kenne zu {\1c&HFF00&\u1}viele{\r} von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.49,0:00:25.51,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.51,0:00:25.65,Default,,0,0,0,,Ich kenne zu viele {\1c&HFF00&\u1}von{\r} ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.65,0:00:25.69,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.69,0:00:25.95,Default,,0,0,0,,Ich kenne zu viele von {\1c&HFF00&\u1}ihren{\r} früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.95,0:00:25.99,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:25.99,0:00:26.41,Default,,0,0,0,,Ich kenne zu viele von ihren {\1c&HFF00&\u1}früheren{\r} Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:26.41,0:00:26.45,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:26.45,0:00:27.01,Default,,0,0,0,,Ich kenne zu viele von ihren früheren {\1c&HFF00&\u1}Eroberungen,{\r} als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.01,0:00:27.06,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.06,0:00:27.24,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, {\1c&HFF00&\u1}als{\r} dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.24,0:00:27.26,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.26,0:00:27.46,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als {\1c&HFF00&\u1}dass{\r} ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.46,0:00:27.50,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.50,0:00:27.60,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass {\1c&HFF00&\u1}ich{\r} in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.60,0:00:27.64,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.64,0:00:27.74,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich {\1c&HFF00&\u1}in{\r} ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.74,0:00:27.78,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:27.78,0:00:28.00,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in {\1c&HFF00&\u1}ihren{\r} Honigtopf treten könnte.
Dialogue: 0,0:00:28.00,0:00:28.04,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:28.04,0:00:28.64,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren {\1c&HFF00&\u1}Honigtopf{\r} treten könnte.
Dialogue: 0,0:00:28.64,0:00:28.68,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:28.68,0:00:28.96,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf {\1c&HFF00&\u1}treten{\r} könnte.
Dialogue: 0,0:00:28.96,0:00:29.00,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
Dialogue: 0,0:00:29.00,0:00:29.24,Default,,0,0,0,,Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten {\1c&HFF00&\u1}könnte.{\r}
Dialogue: 0,0:00:29.24,0:00:29.30,Default,,0,0,0,,{\1c&HFF00&\u1}Na{\r} im Ernst, was ist passiert?
Dialogue: 0,0:00:29.30,0:00:29.32,Default,,0,0,0,,Na im Ernst, was ist passiert?
Dialogue: 0,0:00:29.32,0:00:29.36,Default,,0,0,0,,Na {\1c&HFF00&\u1}im{\r} Ernst, was ist passiert?
Dialogue: 0,0:00:29.36,0:00:29.38,Default,,0,0,0,,Na im Ernst, was ist passiert?
Dialogue: 0,0:00:29.38,0:00:29.48,Default,,0,0,0,,Na im {\1c&HFF00&\u1}Ernst,{\r} was ist passiert?
Dialogue: 0,0:00:29.48,0:00:29.50,Default,,0,0,0,,Na im Ernst, was ist passiert?
Dialogue: 0,0:00:29.50,0:00:29.56,Default,,0,0,0,,Na im Ernst, {\1c&HFF00&\u1}was{\r} ist passiert?
Dialogue: 0,0:00:29.56,0:00:29.58,Default,,0,0,0,,Na im Ernst, was ist passiert?
Dialogue: 0,0:00:29.58,0:00:29.64,Default,,0,0,0,,Na im Ernst, was {\1c&HFF00&\u1}ist{\r} passiert?
Dialogue: 0,0:00:29.64,0:00:29.66,Default,,0,0,0,,Na im Ernst, was ist passiert?
Dialogue: 0,0:00:29.66,0:00:29.82,Default,,0,0,0,,Na im Ernst, was ist {\1c&HFF00&\u1}passiert?{\r}
Dialogue: 0,0:00:30.78,0:00:32.27,Default,,0,0,0,,{\1c&HFF00&\u1}Tja,{\r} ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:32.27,0:00:32.33,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:32.33,0:00:32.49,Default,,0,0,0,,Tja, {\1c&HFF00&\u1}ich{\r} habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:32.49,0:00:32.53,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:32.53,0:00:32.79,Default,,0,0,0,,Tja, ich {\1c&HFF00&\u1}habe{\r} mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:32.79,0:00:32.83,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:32.83,0:00:33.83,Default,,0,0,0,,Tja, ich habe {\1c&HFF00&\u1}mich,{\r} dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:33.83,0:00:33.85,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:33.85,0:00:34.57,Default,,0,0,0,,Tja, ich habe mich, {\1c&HFF00&\u1}dummerweise{\r} muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:34.57,0:00:34.59,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:34.59,0:00:34.73,Default,,0,0,0,,Tja, ich habe mich, dummerweise {\1c&HFF00&\u1}muss{\r} ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:34.73,0:00:34.77,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:34.77,0:00:34.89,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss {\1c&HFF00&\u1}ich{\r} eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:34.89,0:00:34.93,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:34.93,0:00:36.83,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich {\1c&HFF00&\u1}eingestehen,{\r} im Bergsteigen versucht.
Dialogue: 0,0:00:36.83,0:00:36.87,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:36.87,0:00:36.99,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, {\1c&HFF00&\u1}im{\r} Bergsteigen versucht.
Dialogue: 0,0:00:36.99,0:00:37.03,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:37.03,0:00:37.76,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im {\1c&HFF00&\u1}Bergsteigen{\r} versucht.
Dialogue: 0,0:00:37.76,0:00:37.78,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
Dialogue: 0,0:00:37.78,0:00:38.22,Default,,0,0,0,,Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen {\1c&HFF00&\u1}versucht.{\r}
Dialogue: 0,0:00:41.23,0:00:41.85,Default,,0,0,0,,{\1c&HFF00&\u1}Bergsteigen?{\r} Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:41.85,0:00:41.87,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:41.87,0:00:42.13,Default,,0,0,0,,Bergsteigen? {\1c&HFF00&\u1}Dabei{\r} haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.13,0:00:42.15,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.15,0:00:42.33,Default,,0,0,0,,Bergsteigen? Dabei {\1c&HFF00&\u1}haben{\r} sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.33,0:00:42.37,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.37,0:00:42.51,Default,,0,0,0,,Bergsteigen? Dabei haben {\1c&HFF00&\u1}sie{\r} ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.51,0:00:42.55,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.55,0:00:42.61,Default,,0,0,0,,Bergsteigen? Dabei haben sie {\1c&HFF00&\u1}ihr{\r} Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.61,0:00:42.63,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.63,0:00:42.77,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr {\1c&HFF00&\u1}Bein{\r} verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.77,0:00:42.81,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:42.81,0:00:43.23,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein {\1c&HFF00&\u1}verletzt{\r} beim Bergsteigen?
Dialogue: 0,0:00:43.23,0:00:43.27,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:43.27,0:00:43.39,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt {\1c&HFF00&\u1}beim{\r} Bergsteigen?
Dialogue: 0,0:00:43.39,0:00:43.45,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
Dialogue: 0,0:00:43.45,0:00:44.13,Default,,0,0,0,,Bergsteigen? Dabei haben sie ihr Bein verletzt beim {\1c&HFF00&\u1}Bergsteigen?{\r}
Dialogue: 0,0:00:44.54,0:00:44.62,Default,,0,0,0,,{\1c&HFF00&\u1}Ob{\r} sie es glauben oder nicht.
Dialogue: 0,0:00:44.62,0:00:44.66,Default,,0,0,0,,Ob sie es glauben oder nicht.
Dialogue: 0,0:00:44.66,0:00:44.72,Default,,0,0,0,,Ob {\1c&HFF00&\u1}sie{\r} es glauben oder nicht.
Dialogue: 0,0:00:44.72,0:00:44.74,Default,,0,0,0,,Ob sie es glauben oder nicht.
Dialogue: 0,0:00:44.74,0:00:44.80,Default,,0,0,0,,Ob sie {\1c&HFF00&\u1}es{\r} glauben oder nicht.
Dialogue: 0,0:00:44.80,0:00:44.82,Default,,0,0,0,,Ob sie es glauben oder nicht.
Dialogue: 0,0:00:44.82,0:00:45.14,Default,,0,0,0,,Ob sie es {\1c&HFF00&\u1}glauben{\r} oder nicht.
Dialogue: 0,0:00:45.14,0:00:45.16,Default,,0,0,0,,Ob sie es glauben oder nicht.
Dialogue: 0,0:00:45.16,0:00:45.46,Default,,0,0,0,,Ob sie es glauben {\1c&HFF00&\u1}oder{\r} nicht.
Dialogue: 0,0:00:45.46,0:00:45.50,Default,,0,0,0,,Ob sie es glauben oder nicht.
Dialogue: 0,0:00:45.50,0:00:45.66,Default,,0,0,0,,Ob sie es glauben oder {\1c&HFF00&\u1}nicht.{\r}

44
examples/whisperx/sample_de_01.wav.srt
View File

@ -1,44 +0,0 @@
1
00:00:00,563 --> 00:00:01,869
Weinlein von Hammersmann
2
00:00:06,187 --> 00:00:08,013
Oberst Lande, es ist lange her
3
00:00:08,013 --> 00:00:08,534
Schneide ich wie eh und je
4
00:00:13,987 --> 00:00:16,234
Also was ist mit Ihrem wunderschönen Bein geschehen?
5
00:00:17,035 --> 00:00:21,218
Ein Nebenprodukt der Arschtritte, die Sie in der deutschen Filmwelt ausleihen, zweifelsohne
6
00:00:22,102 --> 00:00:24,051
Sparen Sie sich Ihre Komplimente, Sie alter Hund
7
00:00:24,572 --> 00:00:29,238
Ich kenne zu viele von ihren früheren Eroberungen, als dass ich in ihren Honigtopf treten könnte.
8
00:00:29,238 --> 00:00:29,821
Na im Ernst, was ist passiert?
9
00:00:30,783 --> 00:00:38,217
Tja, ich habe mich, dummerweise muss ich eingestehen, im Bergsteigen versucht.
10
00:00:41,226 --> 00:00:44,135
Bergsteigen? Dabei haben sie ihr Bein verletzt beim Bergsteigen?
11
00:00:44,535 --> 00:00:45,657
Ob sie es glauben oder nicht.

199
examples/whisperx/sample_fr_01.wav.ass
View File

@ -1,199 +0,0 @@
[Script Info]
ScriptType: v4.00+
PlayResX: 384
PlayResY: 288
ScaledBorderAndShadow: yes
[V4+ Styles]
Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
Style: Default,Arial,24,&Hffffff,&Hffffff,&H0,&H0,0,0,0,0,100,100,0,0,1,1,0,2,10,10,10,0
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:0.77,0:00:1.07,Default,,0,0,0,,{\1c&HFF00&\u1}Lâchez,{\r} c'est bon.
Dialogue: 0,0:00:1.07,0:00:1.11,Default,,0,0,0,,Lâchez, c'est bon.
Dialogue: 0,0:00:1.11,0:00:1.21,Default,,0,0,0,,Lâchez, {\1c&HFF00&\u1}c'est{\r} bon.
Dialogue: 0,0:00:1.21,0:00:1.23,Default,,0,0,0,,Lâchez, c'est bon.
Dialogue: 0,0:00:1.23,0:00:1.31,Default,,0,0,0,,Lâchez, c'est {\1c&HFF00&\u1}bon.{\r}
Dialogue: 0,0:00:1.31,0:00:1.81,Default,,0,0,0,,{\1c&HFF00&\u1}Ça{\r} va?
Dialogue: 0,0:00:1.81,0:00:1.83,Default,,0,0,0,,Ça va?
Dialogue: 0,0:00:1.83,0:00:1.89,Default,,0,0,0,,Ça {\1c&HFF00&\u1}va?{\r}
Dialogue: 0,0:00:1.91,0:00:1.97,Default,,0,0,0,,{\1c&HFF00&\u1}Oui.{\r}
Dialogue: 0,0:00:1.97,0:00:3.27,Default,,0,0,0,,{\1c&HFF00&\u1}Merci{\r} beaucoup.
Dialogue: 0,0:00:3.27,0:00:3.29,Default,,0,0,0,,Merci beaucoup.
Dialogue: 0,0:00:3.29,0:00:3.49,Default,,0,0,0,,Merci {\1c&HFF00&\u1}beaucoup.{\r}
Dialogue: 0,0:00:4.36,0:00:4.58,Default,,0,0,0,,{\1c&HFF00&\u1}Chèque{\r} ou espèce?
Dialogue: 0,0:00:4.58,0:00:4.64,Default,,0,0,0,,Chèque ou espèce?
Dialogue: 0,0:00:4.64,0:00:4.72,Default,,0,0,0,,Chèque {\1c&HFF00&\u1}ou{\r} espèce?
Dialogue: 0,0:00:4.72,0:00:4.78,Default,,0,0,0,,Chèque ou espèce?
Dialogue: 0,0:00:4.78,0:00:5.04,Default,,0,0,0,,Chèque ou {\1c&HFF00&\u1}espèce?{\r}
Dialogue: 0,0:00:6.54,0:00:6.70,Default,,0,0,0,,{\1c&HFF00&\u1}J'ai{\r} laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:6.70,0:00:6.74,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:6.74,0:00:6.99,Default,,0,0,0,,J'ai {\1c&HFF00&\u1}laissé{\r} un chèque sur la commode, il est signé.
Dialogue: 0,0:00:6.99,0:00:7.03,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:7.03,0:00:7.09,Default,,0,0,0,,J'ai laissé {\1c&HFF00&\u1}un{\r} chèque sur la commode, il est signé.
Dialogue: 0,0:00:7.09,0:00:7.13,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:7.13,0:00:7.33,Default,,0,0,0,,J'ai laissé un {\1c&HFF00&\u1}chèque{\r} sur la commode, il est signé.
Dialogue: 0,0:00:7.33,0:00:7.35,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:7.35,0:00:7.49,Default,,0,0,0,,J'ai laissé un chèque {\1c&HFF00&\u1}sur{\r} la commode, il est signé.
Dialogue: 0,0:00:7.49,0:00:7.51,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:7.51,0:00:7.59,Default,,0,0,0,,J'ai laissé un chèque sur {\1c&HFF00&\u1}la{\r} commode, il est signé.
Dialogue: 0,0:00:7.59,0:00:7.63,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:7.63,0:00:7.91,Default,,0,0,0,,J'ai laissé un chèque sur la {\1c&HFF00&\u1}commode,{\r} il est signé.
Dialogue: 0,0:00:7.91,0:00:7.99,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:7.99,0:00:8.03,Default,,0,0,0,,J'ai laissé un chèque sur la commode, {\1c&HFF00&\u1}il{\r} est signé.
Dialogue: 0,0:00:8.03,0:00:8.09,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:8.09,0:00:8.19,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il {\1c&HFF00&\u1}est{\r} signé.
Dialogue: 0,0:00:8.19,0:00:8.21,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est signé.
Dialogue: 0,0:00:8.21,0:00:8.39,Default,,0,0,0,,J'ai laissé un chèque sur la commode, il est {\1c&HFF00&\u1}signé.{\r}
Dialogue: 0,0:00:8.39,0:00:8.81,Default,,0,0,0,,{\1c&HFF00&\u1}Vous{\r} le remplirez.
Dialogue: 0,0:00:8.81,0:00:8.83,Default,,0,0,0,,Vous le remplirez.
Dialogue: 0,0:00:8.83,0:00:8.95,Default,,0,0,0,,Vous {\1c&HFF00&\u1}le{\r} remplirez.
Dialogue: 0,0:00:8.95,0:00:8.97,Default,,0,0,0,,Vous le remplirez.
Dialogue: 0,0:00:8.97,0:00:9.26,Default,,0,0,0,,Vous le {\1c&HFF00&\u1}remplirez.{\r}
Dialogue: 0,0:00:9.28,0:00:9.34,Default,,0,0,0,,{\1c&HFF00&\u1}OK.{\r}
Dialogue: 0,0:00:9.36,0:00:9.40,Default,,0,0,0,,{\1c&HFF00&\u1}Oh!{\r}
Dialogue: 0,0:00:12.41,0:00:12.51,Default,,0,0,0,,{\1c&HFF00&\u1}Ouh{\r} là!
Dialogue: 0,0:00:12.51,0:00:12.53,Default,,0,0,0,,Ouh là!
Dialogue: 0,0:00:12.53,0:00:12.59,Default,,0,0,0,,Ouh {\1c&HFF00&\u1}là!{\r}
Dialogue: 0,0:00:12.59,0:00:12.73,Default,,0,0,0,,{\1c&HFF00&\u1}Venez.{\r}
Dialogue: 0,0:00:14.45,0:00:14.63,Default,,0,0,0,,{\1c&HFF00&\u1}Merci.{\r}
Dialogue: 0,0:00:14.65,0:00:14.76,Default,,0,0,0,,{\1c&HFF00&\u1}Ah!{\r}
Dialogue: 0,0:00:15.64,0:00:16.51,Default,,0,0,0,,{\1c&HFF00&\u1}C'est{\r} qui?
Dialogue: 0,0:00:16.51,0:00:16.53,Default,,0,0,0,,C'est qui?
Dialogue: 0,0:00:16.53,0:00:16.63,Default,,0,0,0,,C'est {\1c&HFF00&\u1}qui?{\r}
Dialogue: 0,0:00:20.00,0:00:22.85,Default,,0,0,0,,{\1c&HFF00&\u1}C'est{\r} pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:22.85,0:00:22.87,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:22.87,0:00:22.93,Default,,0,0,0,,C'est {\1c&HFF00&\u1}pas{\r} vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:22.93,0:00:22.95,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:22.95,0:00:23.15,Default,,0,0,0,,C'est pas {\1c&HFF00&\u1}vrai,{\r} qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:23.15,0:00:23.17,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:23.17,0:00:23.35,Default,,0,0,0,,C'est pas vrai, {\1c&HFF00&\u1}qu'est-ce{\r} qu'il fout ici, ce con?
Dialogue: 0,0:00:23.35,0:00:23.37,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:23.37,0:00:23.51,Default,,0,0,0,,C'est pas vrai, qu'est-ce {\1c&HFF00&\u1}qu'il{\r} fout ici, ce con?
Dialogue: 0,0:00:23.51,0:00:23.53,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:23.53,0:00:23.67,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il {\1c&HFF00&\u1}fout{\r} ici, ce con?
Dialogue: 0,0:00:23.67,0:00:23.73,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:23.73,0:00:23.95,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout {\1c&HFF00&\u1}ici,{\r} ce con?
Dialogue: 0,0:00:23.95,0:00:23.99,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:23.99,0:00:24.11,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, {\1c&HFF00&\u1}ce{\r} con?
Dialogue: 0,0:00:24.11,0:00:24.15,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
Dialogue: 0,0:00:24.15,0:00:24.23,Default,,0,0,0,,C'est pas vrai, qu'est-ce qu'il fout ici, ce {\1c&HFF00&\u1}con?{\r}
Dialogue: 0,0:00:24.51,0:00:24.96,Default,,0,0,0,,{\1c&HFF00&\u1}Excusez-moi,{\r} mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:24.96,0:00:24.98,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:24.98,0:00:25.06,Default,,0,0,0,,Excusez-moi, {\1c&HFF00&\u1}mais{\r} je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.06,0:00:25.08,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.08,0:00:25.12,Default,,0,0,0,,Excusez-moi, mais {\1c&HFF00&\u1}je{\r} crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.12,0:00:25.14,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.14,0:00:25.26,Default,,0,0,0,,Excusez-moi, mais je {\1c&HFF00&\u1}crois{\r} que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.26,0:00:25.28,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.28,0:00:25.34,Default,,0,0,0,,Excusez-moi, mais je crois {\1c&HFF00&\u1}que{\r} j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.34,0:00:25.36,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.36,0:00:25.42,Default,,0,0,0,,Excusez-moi, mais je crois que {\1c&HFF00&\u1}j'ai{\r} oublié mon sac chez vous.
Dialogue: 0,0:00:25.42,0:00:25.44,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.44,0:00:25.60,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai {\1c&HFF00&\u1}oublié{\r} mon sac chez vous.
Dialogue: 0,0:00:25.60,0:00:25.62,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.62,0:00:25.76,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié {\1c&HFF00&\u1}mon{\r} sac chez vous.
Dialogue: 0,0:00:25.76,0:00:25.78,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.78,0:00:25.94,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon {\1c&HFF00&\u1}sac{\r} chez vous.
Dialogue: 0,0:00:25.94,0:00:25.96,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:25.96,0:00:26.04,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac {\1c&HFF00&\u1}chez{\r} vous.
Dialogue: 0,0:00:26.04,0:00:26.06,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
Dialogue: 0,0:00:26.06,0:00:26.18,Default,,0,0,0,,Excusez-moi, mais je crois que j'ai oublié mon sac chez {\1c&HFF00&\u1}vous.{\r}
Dialogue: 0,0:00:26.18,0:00:26.30,Default,,0,0,0,,{\1c&HFF00&\u1}Ça{\r} va?
Dialogue: 0,0:00:26.30,0:00:26.32,Default,,0,0,0,,Ça va?
Dialogue: 0,0:00:26.32,0:00:26.36,Default,,0,0,0,,Ça {\1c&HFF00&\u1}va?{\r}
Dialogue: 0,0:00:31.04,0:00:31.24,Default,,0,0,0,,{\1c&HFF00&\u1}Attendez.{\r}
Dialogue: 0,0:00:36.81,0:00:36.97,Default,,0,0,0,,{\1c&HFF00&\u1}Tout{\r} à l'heure, là, c'était...
Dialogue: 0,0:00:36.97,0:00:37.01,Default,,0,0,0,,Tout à l'heure, là, c'était...
Dialogue: 0,0:00:37.01,0:00:37.05,Default,,0,0,0,,Tout {\1c&HFF00&\u1}à{\r} l'heure, là, c'était...
Dialogue: 0,0:00:37.05,0:00:37.09,Default,,0,0,0,,Tout à l'heure, là, c'était...
Dialogue: 0,0:00:37.09,0:00:37.35,Default,,0,0,0,,Tout à {\1c&HFF00&\u1}l'heure,{\r} là, c'était...
Dialogue: 0,0:00:37.35,0:00:37.39,Default,,0,0,0,,Tout à l'heure, là, c'était...
Dialogue: 0,0:00:37.39,0:00:37.98,Default,,0,0,0,,Tout à l'heure, {\1c&HFF00&\u1}là,{\r} c'était...
Dialogue: 0,0:00:37.98,0:00:38.02,Default,,0,0,0,,Tout à l'heure, là, c'était...
Dialogue: 0,0:00:38.02,0:00:38.28,Default,,0,0,0,,Tout à l'heure, là, {\1c&HFF00&\u1}c'était...{\r}
Dialogue: 0,0:00:38.28,0:00:38.36,Default,,0,0,0,,{\1c&HFF00&\u1}Vous?{\r}
Dialogue: 0,0:00:39.12,0:00:39.24,Default,,0,0,0,,{\1c&HFF00&\u1}Vous?{\r} Pas...
Dialogue: 0,0:00:39.24,0:00:39.30,Default,,0,0,0,,Vous? Pas...
Dialogue: 0,0:00:39.30,0:00:39.42,Default,,0,0,0,,Vous? {\1c&HFF00&\u1}Pas...{\r}
Dialogue: 0,0:00:39.42,0:00:39.53,Default,,0,0,0,,{\1c&HFF00&\u1}Pas{\r} lui? Vous?
Dialogue: 0,0:00:39.53,0:00:39.55,Default,,0,0,0,,Pas lui? Vous?
Dialogue: 0,0:00:39.55,0:00:39.61,Default,,0,0,0,,Pas {\1c&HFF00&\u1}lui?{\r} Vous?
Dialogue: 0,0:00:39.61,0:00:39.63,Default,,0,0,0,,Pas lui? Vous?
Dialogue: 0,0:00:39.63,0:00:39.71,Default,,0,0,0,,Pas lui? {\1c&HFF00&\u1}Vous?{\r}
Dialogue: 0,0:00:44.19,0:00:44.35,Default,,0,0,0,,{\1c&HFF00&\u1}Vous{\r} avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:44.35,0:00:44.39,Default,,0,0,0,,Vous avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:44.39,0:00:44.62,Default,,0,0,0,,Vous {\1c&HFF00&\u1}avez{\r} tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:44.62,0:00:44.64,Default,,0,0,0,,Vous avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:44.64,0:00:44.80,Default,,0,0,0,,Vous avez {\1c&HFF00&\u1}tout{\r} à fait raison, M. Xenakis.
Dialogue: 0,0:00:44.80,0:00:44.82,Default,,0,0,0,,Vous avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:44.82,0:00:44.84,Default,,0,0,0,,Vous avez tout {\1c&HFF00&\u1}à{\r} fait raison, M. Xenakis.
Dialogue: 0,0:00:44.84,0:00:44.90,Default,,0,0,0,,Vous avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:44.90,0:00:45.04,Default,,0,0,0,,Vous avez tout à {\1c&HFF00&\u1}fait{\r} raison, M. Xenakis.
Dialogue: 0,0:00:45.04,0:00:45.08,Default,,0,0,0,,Vous avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:45.08,0:00:45.38,Default,,0,0,0,,Vous avez tout à fait {\1c&HFF00&\u1}raison,{\r} M. Xenakis.
Dialogue: 0,0:00:45.38,0:00:45.42,Default,,0,0,0,,Vous avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:45.42,0:00:45.62,Default,,0,0,0,,Vous avez tout à fait raison, {\1c&HFF00&\u1}M.{\r} Xenakis.
Dialogue: 0,0:00:45.62,0:00:45.68,Default,,0,0,0,,Vous avez tout à fait raison, M. Xenakis.
Dialogue: 0,0:00:45.68,0:00:45.98,Default,,0,0,0,,Vous avez tout à fait raison, M. {\1c&HFF00&\u1}Xenakis.{\r}
Dialogue: 0,0:00:46.75,0:00:47.13,Default,,0,0,0,,{\1c&HFF00&\u1}Malek{\r} est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.13,0:00:47.15,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.15,0:00:47.27,Default,,0,0,0,,Malek {\1c&HFF00&\u1}est{\r} à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.27,0:00:47.31,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.31,0:00:47.37,Default,,0,0,0,,Malek est {\1c&HFF00&\u1}à{\r} l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.37,0:00:47.39,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.39,0:00:47.75,Default,,0,0,0,,Malek est à {\1c&HFF00&\u1}l'interne{\r} brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.75,0:00:47.79,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:47.79,0:00:48.07,Default,,0,0,0,,Malek est à l'interne {\1c&HFF00&\u1}brillant,{\r} qui apprend le métier avec moi.
Dialogue: 0,0:00:48.07,0:00:48.11,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:48.11,0:00:48.19,Default,,0,0,0,,Malek est à l'interne brillant, {\1c&HFF00&\u1}qui{\r} apprend le métier avec moi.
Dialogue: 0,0:00:48.19,0:00:48.23,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:48.23,0:00:48.44,Default,,0,0,0,,Malek est à l'interne brillant, qui {\1c&HFF00&\u1}apprend{\r} le métier avec moi.
Dialogue: 0,0:00:48.44,0:00:48.46,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:48.46,0:00:48.52,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend {\1c&HFF00&\u1}le{\r} métier avec moi.
Dialogue: 0,0:00:48.52,0:00:48.54,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:48.54,0:00:48.74,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le {\1c&HFF00&\u1}métier{\r} avec moi.
Dialogue: 0,0:00:48.74,0:00:48.76,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:48.76,0:00:48.88,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier {\1c&HFF00&\u1}avec{\r} moi.
Dialogue: 0,0:00:48.88,0:00:48.90,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec moi.
Dialogue: 0,0:00:48.90,0:00:49.00,Default,,0,0,0,,Malek est à l'interne brillant, qui apprend le métier avec {\1c&HFF00&\u1}moi.{\r}
Dialogue: 0,0:00:49.02,0:00:49.06,Default,,0,0,0,,{\1c&HFF00&\u1}Ah!{\r}
Dialogue: 0,0:00:49.06,0:00:49.20,Default,,0,0,0,,{\1c&HFF00&\u1}C'est{\r} vrai.
Dialogue: 0,0:00:49.20,0:00:49.22,Default,,0,0,0,,C'est vrai.
Dialogue: 0,0:00:49.22,0:00:49.30,Default,,0,0,0,,C'est {\1c&HFF00&\u1}vrai.{\r}
Dialogue: 0,0:00:49.30,0:00:49.44,Default,,0,0,0,,{\1c&HFF00&\u1}Bien.{\r}
Dialogue: 0,0:00:52.93,0:00:53.21,Default,,0,0,0,,{\1c&HFF00&\u1}Justement,{\r} y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.21,0:00:53.33,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.33,0:00:53.35,Default,,0,0,0,,Justement, {\1c&HFF00&\u1}y{\r} a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.35,0:00:53.37,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.37,0:00:53.39,Default,,0,0,0,,Justement, y {\1c&HFF00&\u1}a{\r} la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.39,0:00:53.43,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.43,0:00:53.49,Default,,0,0,0,,Justement, y a {\1c&HFF00&\u1}la{\r} famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.49,0:00:53.51,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.51,0:00:53.85,Default,,0,0,0,,Justement, y a la {\1c&HFF00&\u1}famille{\r} Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.85,0:00:53.89,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:53.89,0:00:54.16,Default,,0,0,0,,Justement, y a la famille {\1c&HFF00&\u1}Boboun{\r} qui m'attend pour une consultation.
Dialogue: 0,0:00:54.16,0:00:54.18,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:54.18,0:00:54.26,Default,,0,0,0,,Justement, y a la famille Boboun {\1c&HFF00&\u1}qui{\r} m'attend pour une consultation.
Dialogue: 0,0:00:54.26,0:00:54.30,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:54.30,0:00:54.52,Default,,0,0,0,,Justement, y a la famille Boboun qui {\1c&HFF00&\u1}m'attend{\r} pour une consultation.
Dialogue: 0,0:00:54.52,0:00:54.54,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:54.54,0:00:54.64,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend {\1c&HFF00&\u1}pour{\r} une consultation.
Dialogue: 0,0:00:54.64,0:00:54.68,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:54.68,0:00:54.90,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour {\1c&HFF00&\u1}une{\r} consultation.
Dialogue: 0,0:00:54.90,0:00:54.94,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une consultation.
Dialogue: 0,0:00:54.94,0:00:55.34,Default,,0,0,0,,Justement, y a la famille Boboun qui m'attend pour une {\1c&HFF00&\u1}consultation.{\r}
Dialogue: 0,0:00:55.58,0:00:55.64,Default,,0,0,0,,{\1c&HFF00&\u1}Qui?{\r}
Dialogue: 0,0:00:56.53,0:00:56.79,Default,,0,0,0,,{\1c&HFF00&\u1}Faisons{\r} pas attendre les bobounes, allez.
Dialogue: 0,0:00:56.79,0:00:56.81,Default,,0,0,0,,Faisons pas attendre les bobounes, allez.
Dialogue: 0,0:00:56.81,0:00:56.91,Default,,0,0,0,,Faisons {\1c&HFF00&\u1}pas{\r} attendre les bobounes, allez.
Dialogue: 0,0:00:56.91,0:00:56.93,Default,,0,0,0,,Faisons pas attendre les bobounes, allez.
Dialogue: 0,0:00:56.93,0:00:57.15,Default,,0,0,0,,Faisons pas {\1c&HFF00&\u1}attendre{\r} les bobounes, allez.
Dialogue: 0,0:00:57.15,0:00:57.19,Default,,0,0,0,,Faisons pas attendre les bobounes, allez.
Dialogue: 0,0:00:57.19,0:00:57.25,Default,,0,0,0,,Faisons pas attendre {\1c&HFF00&\u1}les{\r} bobounes, allez.
Dialogue: 0,0:00:57.25,0:00:57.27,Default,,0,0,0,,Faisons pas attendre les bobounes, allez.
Dialogue: 0,0:00:57.27,0:00:57.59,Default,,0,0,0,,Faisons pas attendre les {\1c&HFF00&\u1}bobounes,{\r} allez.
Dialogue: 0,0:00:57.59,0:00:57.61,Default,,0,0,0,,Faisons pas attendre les bobounes, allez.
Dialogue: 0,0:00:57.61,0:00:57.75,Default,,0,0,0,,Faisons pas attendre les bobounes, {\1c&HFF00&\u1}allez.{\r}

120
examples/whisperx/sample_fr_01.wav.srt
View File

@ -1,120 +0,0 @@
1
00:00:00,765 --> 00:00:01,309
Lâchez, c'est bon.
2
00:00:01,309 --> 00:00:01,891
Ça va?
3
00:00:01,911 --> 00:00:01,971
Oui.
4
00:00:01,971 --> 00:00:03,495
Merci beaucoup.
5
00:00:04,356 --> 00:00:05,037
Chèque ou espèce?
6
00:00:06,544 --> 00:00:08,393
J'ai laissé un chèque sur la commode, il est signé.
7
00:00:08,393 --> 00:00:09,255
Vous le remplirez.
8
00:00:09,275 --> 00:00:09,335
OK.
9
00:00:09,355 --> 00:00:09,395
Oh!
10
00:00:12,410 --> 00:00:12,590
Ouh là!
11
00:00:12,590 --> 00:00:12,731
Venez.
12
00:00:14,454 --> 00:00:14,635
Merci.
13
00:00:14,655 --> 00:00:14,755
Ah!
14
00:00:15,640 --> 00:00:16,626
C'est qui?
15
00:00:20,000 --> 00:00:24,234
C'est pas vrai, qu'est-ce qu'il fout ici, ce con?
16
00:00:24,515 --> 00:00:26,177
Excusez-moi, mais je crois que j'ai oublié mon sac chez vous.
17
00:00:26,177 --> 00:00:26,359
Ça va?
18
00:00:31,040 --> 00:00:31,241
Attendez.
19
00:00:36,813 --> 00:00:38,278
Tout à l'heure, là, c'était...
20
00:00:38,278 --> 00:00:38,359
Vous?
21
00:00:39,123 --> 00:00:39,425
Vous? Pas...
22
00:00:39,425 --> 00:00:39,706
Pas lui? Vous?
23
00:00:44,194 --> 00:00:45,980
Vous avez tout à fait raison, M. Xenakis.
24
00:00:46,745 --> 00:00:49,000
Malek est à l'interne brillant, qui apprend le métier avec moi.
25
00:00:49,020 --> 00:00:49,061
Ah!
26
00:00:49,061 --> 00:00:49,303
C'est vrai.
27
00:00:49,303 --> 00:00:49,443
Bien.
28
00:00:52,932 --> 00:00:55,338
Justement, y a la famille Boboun qui m'attend pour une consultation.
29
00:00:55,581 --> 00:00:55,642
Qui?
30
00:00:56,527 --> 00:00:57,753
Faisons pas attendre les bobounes, allez.

177
examples/whisperx/sample_it_01.wav.ass
View File

@ -1,177 +0,0 @@
[Script Info]
ScriptType: v4.00+
PlayResX: 384
PlayResY: 288
ScaledBorderAndShadow: yes
[V4+ Styles]
Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
Style: Default,Arial,24,&Hffffff,&Hffffff,&H0,&H0,0,0,0,0,100,100,0,0,1,1,0,2,10,10,10,0
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:1.20,0:00:1.62,Default,,0,0,0,,{\1c&HFF00&\u1}Signore,{\r} è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:1.62,0:00:1.64,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:1.64,0:00:1.66,Default,,0,0,0,,Signore, {\1c&HFF00&\u1}è{\r} un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:1.66,0:00:1.72,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:1.72,0:00:2.12,Default,,0,0,0,,Signore, è {\1c&HFF00&\u1}un{\r} piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:2.12,0:00:2.18,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:2.18,0:00:2.72,Default,,0,0,0,,Signore, è un {\1c&HFF00&\u1}piacere,{\r} gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:2.72,0:00:3.33,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:3.33,0:00:3.45,Default,,0,0,0,,Signore, è un piacere, {\1c&HFF00&\u1}gli{\r} amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:3.45,0:00:3.49,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:3.49,0:00:3.79,Default,,0,0,0,,Signore, è un piacere, gli {\1c&HFF00&\u1}amici{\r} della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:3.79,0:00:3.83,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:3.83,0:00:4.01,Default,,0,0,0,,Signore, è un piacere, gli amici {\1c&HFF00&\u1}della{\r} vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.01,0:00:4.05,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.05,0:00:4.35,Default,,0,0,0,,Signore, è un piacere, gli amici della {\1c&HFF00&\u1}vedetta{\r} ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.35,0:00:4.39,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.39,0:00:4.79,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta {\1c&HFF00&\u1}ammirata{\r} da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.79,0:00:4.85,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.85,0:00:4.95,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata {\1c&HFF00&\u1}da{\r} tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.95,0:00:4.97,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:4.97,0:00:5.15,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da {\1c&HFF00&\u1}tutti{\r} noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:5.15,0:00:5.21,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:5.21,0:00:5.33,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti {\1c&HFF00&\u1}noi,{\r} questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:5.33,0:00:5.41,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:5.41,0:00:5.61,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, {\1c&HFF00&\u1}questa{\r} gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:5.61,0:00:5.79,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:5.79,0:00:6.07,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa {\1c&HFF00&\u1}gemma{\r} propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:6.07,0:00:6.13,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:6.13,0:00:6.51,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma {\1c&HFF00&\u1}propria{\r} della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:6.51,0:00:6.57,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:6.57,0:00:6.77,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria {\1c&HFF00&\u1}della{\r} nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:6.77,0:00:6.81,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:6.81,0:00:6.99,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della {\1c&HFF00&\u1}nostra{\r} cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:6.99,0:00:7.07,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:7.07,0:00:7.35,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra {\1c&HFF00&\u1}cultura,{\r} saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:7.35,0:00:7.41,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:7.41,0:00:7.73,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, {\1c&HFF00&\u1}saranno{\r} naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:7.73,0:00:7.87,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:7.87,0:00:8.47,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno {\1c&HFF00&\u1}naturalmente{\r} accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:8.47,0:00:8.55,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:8.55,0:00:8.85,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente {\1c&HFF00&\u1}accolti{\r} sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:8.85,0:00:8.91,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:8.91,0:00:9.07,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti {\1c&HFF00&\u1}sotto{\r} la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:9.07,0:00:9.13,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:9.13,0:00:9.19,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto {\1c&HFF00&\u1}la{\r} mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:9.19,0:00:9.23,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:9.23,0:00:9.33,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la {\1c&HFF00&\u1}mia{\r} protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:9.33,0:00:9.37,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:9.37,0:00:9.82,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia {\1c&HFF00&\u1}protezione{\r} per la durata del loro soggiorno.
Dialogue: 0,0:00:9.82,0:00:9.88,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:9.88,0:00:9.96,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione {\1c&HFF00&\u1}per{\r} la durata del loro soggiorno.
Dialogue: 0,0:00:9.96,0:00:10.02,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:10.02,0:00:10.08,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per {\1c&HFF00&\u1}la{\r} durata del loro soggiorno.
Dialogue: 0,0:00:10.08,0:00:10.12,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:10.12,0:00:10.44,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la {\1c&HFF00&\u1}durata{\r} del loro soggiorno.
Dialogue: 0,0:00:10.44,0:00:10.50,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:10.50,0:00:10.60,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata {\1c&HFF00&\u1}del{\r} loro soggiorno.
Dialogue: 0,0:00:10.60,0:00:10.62,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:10.62,0:00:10.78,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del {\1c&HFF00&\u1}loro{\r} soggiorno.
Dialogue: 0,0:00:10.78,0:00:10.86,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
Dialogue: 0,0:00:10.86,0:00:11.30,Default,,0,0,0,,Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro {\1c&HFF00&\u1}soggiorno.{\r}
Dialogue: 0,0:00:13.00,0:00:13.12,Default,,0,0,0,,{\1c&HFF00&\u1}Grazie.{\r}
Dialogue: 0,0:00:15.60,0:00:17.89,Default,,0,0,0,,{\1c&HFF00&\u1}Gorlami?{\r} Lo pronuncio correttamente?
Dialogue: 0,0:00:17.89,0:00:18.75,Default,,0,0,0,,Gorlami? Lo pronuncio correttamente?
Dialogue: 0,0:00:18.75,0:00:18.83,Default,,0,0,0,,Gorlami? {\1c&HFF00&\u1}Lo{\r} pronuncio correttamente?
Dialogue: 0,0:00:18.83,0:00:18.87,Default,,0,0,0,,Gorlami? Lo pronuncio correttamente?
Dialogue: 0,0:00:18.87,0:00:19.31,Default,,0,0,0,,Gorlami? Lo {\1c&HFF00&\u1}pronuncio{\r} correttamente?
Dialogue: 0,0:00:19.31,0:00:19.37,Default,,0,0,0,,Gorlami? Lo pronuncio correttamente?
Dialogue: 0,0:00:19.37,0:00:19.87,Default,,0,0,0,,Gorlami? Lo pronuncio {\1c&HFF00&\u1}correttamente?{\r}
Dialogue: 0,0:00:21.58,0:00:21.74,Default,,0,0,0,,{\1c&HFF00&\u1}Sì,{\r} corretto.
Dialogue: 0,0:00:21.74,0:00:22.34,Default,,0,0,0,,Sì, corretto.
Dialogue: 0,0:00:22.34,0:00:22.72,Default,,0,0,0,,Sì, {\1c&HFF00&\u1}corretto.{\r}
Dialogue: 0,0:00:23.54,0:00:24.82,Default,,0,0,0,,{\1c&HFF00&\u1}Gorlami?{\r} Per cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:24.82,0:00:25.45,Default,,0,0,0,,Gorlami? Per cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:25.45,0:00:25.53,Default,,0,0,0,,Gorlami? {\1c&HFF00&\u1}Per{\r} cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:25.53,0:00:25.61,Default,,0,0,0,,Gorlami? Per cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:25.61,0:00:26.17,Default,,0,0,0,,Gorlami? Per {\1c&HFF00&\u1}cortesia,{\r} me lo ripeto ancora.
Dialogue: 0,0:00:26.17,0:00:26.39,Default,,0,0,0,,Gorlami? Per cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:26.39,0:00:26.45,Default,,0,0,0,,Gorlami? Per cortesia, {\1c&HFF00&\u1}me{\r} lo ripeto ancora.
Dialogue: 0,0:00:26.45,0:00:26.49,Default,,0,0,0,,Gorlami? Per cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:26.49,0:00:26.55,Default,,0,0,0,,Gorlami? Per cortesia, me {\1c&HFF00&\u1}lo{\r} ripeto ancora.
Dialogue: 0,0:00:26.55,0:00:26.61,Default,,0,0,0,,Gorlami? Per cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:26.61,0:00:27.05,Default,,0,0,0,,Gorlami? Per cortesia, me lo {\1c&HFF00&\u1}ripeto{\r} ancora.
Dialogue: 0,0:00:27.05,0:00:27.11,Default,,0,0,0,,Gorlami? Per cortesia, me lo ripeto ancora.
Dialogue: 0,0:00:27.11,0:00:27.49,Default,,0,0,0,,Gorlami? Per cortesia, me lo ripeto {\1c&HFF00&\u1}ancora.{\r}
Dialogue: 0,0:00:27.55,0:00:28.79,Default,,0,0,0,,{\1c&HFF00&\u1}ancora{\r} gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:28.79,0:00:28.83,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:28.83,0:00:30.96,Default,,0,0,0,,ancora {\1c&HFF00&\u1}gourlami{\r} scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:30.96,0:00:31.02,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:31.02,0:00:31.36,Default,,0,0,0,,ancora gourlami {\1c&HFF00&\u1}scusi{\r} con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:31.36,0:00:31.46,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:31.46,0:00:31.66,Default,,0,0,0,,ancora gourlami scusi {\1c&HFF00&\u1}con{\r} me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:31.66,0:00:31.72,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:31.72,0:00:31.88,Default,,0,0,0,,ancora gourlami scusi con {\1c&HFF00&\u1}me{\r} gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:31.88,0:00:33.46,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:33.46,0:00:34.12,Default,,0,0,0,,ancora gourlami scusi con me {\1c&HFF00&\u1}gourlami{\r} ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:34.12,0:00:34.84,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:34.84,0:00:35.20,Default,,0,0,0,,ancora gourlami scusi con me gourlami {\1c&HFF00&\u1}ancora{\r} una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:35.20,0:00:35.32,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:35.32,0:00:35.44,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora {\1c&HFF00&\u1}una{\r} volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:35.44,0:00:35.48,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:35.48,0:00:35.82,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una {\1c&HFF00&\u1}volta{\r} gourlami e come si chiama lei antonio
Dialogue: 0,0:00:35.82,0:00:39.16,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:39.16,0:00:39.72,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta {\1c&HFF00&\u1}gourlami{\r} e come si chiama lei antonio
Dialogue: 0,0:00:39.72,0:00:40.96,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:40.96,0:00:41.14,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami {\1c&HFF00&\u1}e{\r} come si chiama lei antonio
Dialogue: 0,0:00:41.14,0:00:41.20,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:41.20,0:00:41.34,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e {\1c&HFF00&\u1}come{\r} si chiama lei antonio
Dialogue: 0,0:00:41.34,0:00:41.38,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:41.38,0:00:41.46,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come {\1c&HFF00&\u1}si{\r} chiama lei antonio
Dialogue: 0,0:00:41.46,0:00:41.50,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:41.50,0:00:41.70,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si {\1c&HFF00&\u1}chiama{\r} lei antonio
Dialogue: 0,0:00:41.70,0:00:41.80,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:41.80,0:00:42.06,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama {\1c&HFF00&\u1}lei{\r} antonio
Dialogue: 0,0:00:42.06,0:00:43.44,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
Dialogue: 0,0:00:43.44,0:00:43.98,Default,,0,0,0,,ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei {\1c&HFF00&\u1}antonio{\r}
Dialogue: 0,0:00:44.36,0:00:45.14,Default,,0,0,0,,{\1c&HFF00&\u1}margarete{\r} ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:45.14,0:00:46.56,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:46.56,0:00:46.85,Default,,0,0,0,,margarete {\1c&HFF00&\u1}ancora{\r} margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:46.85,0:00:47.85,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:47.85,0:00:49.15,Default,,0,0,0,,margarete ancora {\1c&HFF00&\u1}margarete{\r} un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:49.15,0:00:49.43,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:49.43,0:00:49.85,Default,,0,0,0,,margarete ancora margarete {\1c&HFF00&\u1}un'altra{\r} volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:49.85,0:00:49.91,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:49.91,0:00:50.13,Default,,0,0,0,,margarete ancora margarete un'altra {\1c&HFF00&\u1}volta{\r} ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:50.13,0:00:50.19,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:50.19,0:00:50.37,Default,,0,0,0,,margarete ancora margarete un'altra volta {\1c&HFF00&\u1}ma{\r} adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:50.37,0:00:50.43,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:50.43,0:00:50.73,Default,,0,0,0,,margarete ancora margarete un'altra volta ma {\1c&HFF00&\u1}adesso{\r} vorrei proprio sentire la musica
Dialogue: 0,0:00:50.73,0:00:50.77,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:50.77,0:00:51.01,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso {\1c&HFF00&\u1}vorrei{\r} proprio sentire la musica
Dialogue: 0,0:00:51.01,0:00:51.05,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:51.05,0:00:51.29,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei {\1c&HFF00&\u1}proprio{\r} sentire la musica
Dialogue: 0,0:00:51.29,0:00:51.43,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:51.43,0:00:51.71,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio {\1c&HFF00&\u1}sentire{\r} la musica
Dialogue: 0,0:00:51.71,0:00:51.79,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:51.79,0:00:51.88,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire {\1c&HFF00&\u1}la{\r} musica
Dialogue: 0,0:00:51.88,0:00:51.92,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
Dialogue: 0,0:00:51.92,0:00:52.36,Default,,0,0,0,,margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la {\1c&HFF00&\u1}musica{\r}
Dialogue: 0,0:00:52.36,0:00:52.40,Default,,0,0,0,,{\1c&HFF00&\u1}le{\r} parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:52.40,0:00:52.52,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:52.52,0:00:54.32,Default,,0,0,0,,le {\1c&HFF00&\u1}parole{\r} margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:54.32,0:00:54.40,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:54.40,0:00:56.40,Default,,0,0,0,,le parole {\1c&HFF00&\u1}margheriti{\r} margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:56.40,0:00:56.96,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:56.96,0:00:57.92,Default,,0,0,0,,le parole margheriti {\1c&HFF00&\u1}margheriti{\r} e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:57.92,0:00:58.53,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:58.53,0:00:58.59,Default,,0,0,0,,le parole margheriti margheriti {\1c&HFF00&\u1}e{\r} lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:58.59,0:00:58.65,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:58.65,0:00:58.97,Default,,0,0,0,,le parole margheriti margheriti e {\1c&HFF00&\u1}lei{\r} dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:58.97,0:00:59.59,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:59.59,0:00:59.91,Default,,0,0,0,,le parole margheriti margheriti e lei {\1c&HFF00&\u1}dominic{\r} decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:59.91,0:00:59.95,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:00:59.95,0:01:0.35,Default,,0,0,0,,le parole margheriti margheriti e lei dominic {\1c&HFF00&\u1}decoco{\r} come dominic decoco bravo bravo
Dialogue: 0,0:01:0.35,0:01:0.55,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:01:0.55,0:01:0.73,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco {\1c&HFF00&\u1}come{\r} dominic decoco bravo bravo
Dialogue: 0,0:01:0.73,0:01:1.25,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:01:1.25,0:01:1.57,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come {\1c&HFF00&\u1}dominic{\r} decoco bravo bravo
Dialogue: 0,0:01:1.57,0:01:1.61,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:01:1.61,0:01:2.01,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic {\1c&HFF00&\u1}decoco{\r} bravo bravo
Dialogue: 0,0:01:2.01,0:01:2.17,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:01:2.17,0:01:2.45,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco {\1c&HFF00&\u1}bravo{\r} bravo
Dialogue: 0,0:01:2.45,0:01:2.91,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo
Dialogue: 0,0:01:2.91,0:01:3.29,Default,,0,0,0,,le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo {\1c&HFF00&\u1}bravo{\r}

32
examples/whisperx/sample_it_01.wav.srt
View File

@ -1,32 +0,0 @@
1
00:00:01,202 --> 00:00:11,297
Signore, è un piacere, gli amici della vedetta ammirata da tutti noi, questa gemma propria della nostra cultura, saranno naturalmente accolti sotto la mia protezione per la durata del loro soggiorno.
2
00:00:13,000 --> 00:00:13,120
Grazie.
3
00:00:15,602 --> 00:00:19,874
Gorlami? Lo pronuncio correttamente?
4
00:00:21,576 --> 00:00:22,717
Sì, corretto.
5
00:00:23,540 --> 00:00:27,495
Gorlami? Per cortesia, me lo ripeto ancora.
6
00:00:27,555 --> 00:00:43,979
ancora gourlami scusi con me gourlami ancora una volta gourlami e come si chiama lei antonio
7
00:00:44,360 --> 00:00:52,356
margarete ancora margarete un'altra volta ma adesso vorrei proprio sentire la musica
8
00:00:52,356 --> 00:01:03,292
le parole margheriti margheriti e lei dominic decoco come dominic decoco bravo bravo

7
requirements.txt
View File

@ -1,9 +1,10 @@
numpy
torch
torchaudio
pandas
torch >=1.9
torchaudio >=0.10,<1.0
tqdm
more-itertools
transformers>=4.19.0
ffmpeg-python==0.2.0
pyannote.audio
soundfile
whisper

2
setup.py
View File

@ -6,7 +6,7 @@ from setuptools import setup, find_packages
setup(
name="whisperx",
py_modules=["whisperx"],
version="1.0",
version="2.0",
description="Time-Accurate Automatic Speech Recognition using Whisper.",
readme="README.md",
python_requires=">=3.7",

BIN
tests/jfk.flac
View File

Binary file not shown.

19
tests/test_audio.py
View File

@ -1,19 +0,0 @@
import os.path
import numpy as np
from whisper.audio import load_audio, log_mel_spectrogram, SAMPLE_RATE
def test_audio():
audio_path = os.path.join(os.path.dirname(__file__), "jfk.flac")
audio = load_audio(audio_path)
assert audio.ndim == 1
assert SAMPLE_RATE * 10 < audio.shape[0] < SAMPLE_RATE * 12
assert 0 < audio.std() < 1
mel_from_audio = log_mel_spectrogram(audio)
mel_from_file = log_mel_spectrogram(audio_path)
assert np.allclose(mel_from_audio, mel_from_file)
assert mel_from_audio.max() - mel_from_audio.min() <= 2.0

92
tests/test_normalizer.py
View File

@ -1,92 +0,0 @@
import pytest
from whisper.normalizers import EnglishTextNormalizer
from whisper.normalizers.english import EnglishNumberNormalizer, EnglishSpellingNormalizer
@pytest.mark.parametrize("std", [EnglishNumberNormalizer(), EnglishTextNormalizer()])
def test_number_normalizer(std):
assert std("two") == "2"
assert std("thirty one") == "31"
assert std("five twenty four") == "524"
assert std("nineteen ninety nine") == "1999"
assert std("twenty nineteen") == "2019"
assert std("two point five million") == "2500000"
assert std("four point two billions") == "4200000000s"
assert std("200 thousand") == "200000"
assert std("200 thousand dollars") == "$200000"
assert std("$20 million") == "$20000000"
assert std("€52.4 million") == "€52400000"
assert std("£77 thousands") == "£77000s"
assert std("two double o eight") == "2008"
assert std("three thousand twenty nine") == "3029"
assert std("forty three thousand two hundred sixty") == "43260"
assert std("forty three thousand two hundred and sixty") == "43260"
assert std("nineteen fifties") == "1950s"
assert std("thirty first") == "31st"
assert std("thirty three thousand and three hundred and thirty third") == "33333rd"
assert std("three billion") == "3000000000"
assert std("millions") == "1000000s"
assert std("july third twenty twenty") == "july 3rd 2020"
assert std("august twenty sixth twenty twenty one") == "august 26th 2021"
assert std("3 14") == "3 14"
assert std("3.14") == "3.14"
assert std("3 point 2") == "3.2"
assert std("3 point 14") == "3.14"
assert std("fourteen point 4") == "14.4"
assert std("two point two five dollars") == "$2.25"
assert std("two hundred million dollars") == "$200000000"
assert std("$20.1 million") == "$20100000"
assert std("ninety percent") == "90%"
assert std("seventy six per cent") == "76%"
assert std("double oh seven") == "007"
assert std("double zero seven") == "007"
assert std("nine one one") == "911"
assert std("nine double one") == "911"
assert std("one triple oh one") == "10001"
assert std("two thousandth") == "2000th"
assert std("thirty two thousandth") == "32000th"
assert std("minus 500") == "-500"
assert std("positive twenty thousand") == "+20000"
assert std("two dollars and seventy cents") == "$2.70"
assert std("3 cents") == "¢3"
assert std("$0.36") == "¢36"
assert std("three euros and sixty five cents") == "€3.65"
assert std("three and a half million") == "3500000"
assert std("forty eight and a half dollars") == "$48.5"
assert std("b747") == "b 747"
assert std("10 th") == "10th"
assert std("10th") == "10th"
def test_spelling_normalizer():
std = EnglishSpellingNormalizer()
assert std("mobilisation") == "mobilization"
assert std("cancelation") == "cancellation"
def test_text_normalizer():
std = EnglishTextNormalizer()
assert std("Let's") == "let us"
assert std("he's like") == "he is like"
assert std("she's been like") == "she has been like"
assert std("10km") == "10 km"
assert std("RC232") == "rc 232"
assert (
std("Mr. Park visited Assoc. Prof. Kim Jr.")
== "mister park visited associate professor kim junior"
)

14
tests/test_tokenizer.py
View File

@ -1,14 +0,0 @@
from whisper.tokenizer import get_tokenizer
def test_tokenizer():
gpt2_tokenizer = get_tokenizer(multilingual=False)
multilingual_tokenizer = get_tokenizer(multilingual=True)
text = "다람쥐 헌 쳇바퀴에 타고파"
gpt2_tokens = gpt2_tokenizer.encode(text)
multilingual_tokens = multilingual_tokenizer.encode(text)
assert gpt2_tokenizer.decode(gpt2_tokens) == text
assert multilingual_tokenizer.decode(multilingual_tokens) == text
assert len(gpt2_tokens) > len(multilingual_tokens)

20
tests/test_transcribe.py
View File

@ -1,20 +0,0 @@
import os
import pytest
import whisper
@pytest.mark.parametrize('model_name', whisper.available_models())
def test_transcribe(model_name: str):
model = whisper.load_model(model_name).cuda()
audio_path = os.path.join(os.path.dirname(__file__), "jfk.flac")
language = "en" if model_name.endswith(".en") else None
result = model.transcribe(audio_path, language=language, temperature=0.0)
assert result["language"] == "en"
transcription = result["text"].lower()
assert "my fellow americans" in transcription
assert "your country" in transcription
assert "do for you" in transcription

116
whisperx/__init__.py
View File

@ -1,115 +1,3 @@
import hashlib
import io
import os
import urllib
import warnings
from typing import List, Optional, Union
import torch
from tqdm import tqdm
from .audio import load_audio, log_mel_spectrogram, pad_or_trim
from .decoding import DecodingOptions, DecodingResult, decode, detect_language
from .model import Whisper, ModelDimensions
from .transcribe import transcribe, transcribe_with_vad, transcribe_with_vad_parallel
from .transcribe import transcribe, transcribe_with_vad
from .alignment import load_align_model, align
_MODELS = {
"tiny.en": "https://openaipublic.azureedge.net/main/whisper/models/d3dd57d32accea0b295c96e26691aa14d8822fac7d9d27d5dc00b4ca2826dd03/tiny.en.pt",
"tiny": "https://openaipublic.azureedge.net/main/whisper/models/65147644a518d12f04e32d6f3b26facc3f8dd46e5390956a9424a650c0ce22b9/tiny.pt",
"base.en": "https://openaipublic.azureedge.net/main/whisper/models/25a8566e1d0c1e2231d1c762132cd20e0f96a85d16145c3a00adf5d1ac670ead/base.en.pt",
"base": "https://openaipublic.azureedge.net/main/whisper/models/ed3a0b6b1c0edf879ad9b11b1af5a0e6ab5db9205f891f668f8b0e6c6326e34e/base.pt",
"small.en": "https://openaipublic.azureedge.net/main/whisper/models/f953ad0fd29cacd07d5a9eda5624af0f6bcf2258be67c92b79389873d91e0872/small.en.pt",
"small": "https://openaipublic.azureedge.net/main/whisper/models/9ecf779972d90ba49c06d968637d720dd632c55bbf19d441fb42bf17a411e794/small.pt",
"medium.en": "https://openaipublic.azureedge.net/main/whisper/models/d7440d1dc186f76616474e0ff0b3b6b879abc9d1a4926b7adfa41db2d497ab4f/medium.en.pt",
"medium": "https://openaipublic.azureedge.net/main/whisper/models/345ae4da62f9b3d59415adc60127b97c714f32e89e936602e85993674d08dcb1/medium.pt",
"large-v1": "https://openaipublic.azureedge.net/main/whisper/models/e4b87e7e0bf463eb8e6956e646f1e277e901512310def2c24bf0e11bd3c28e9a/large-v1.pt",
"large-v2": "https://openaipublic.azureedge.net/main/whisper/models/81f7c96c852ee8fc832187b0132e569d6c3065a3252ed18e56effd0b6a73e524/large-v2.pt",
"large": "https://openaipublic.azureedge.net/main/whisper/models/81f7c96c852ee8fc832187b0132e569d6c3065a3252ed18e56effd0b6a73e524/large-v2.pt",
}
def _download(url: str, root: str, in_memory: bool) -> Union[bytes, str]:
os.makedirs(root, exist_ok=True)
expected_sha256 = url.split("/")[-2]
download_target = os.path.join(root, os.path.basename(url))
if os.path.exists(download_target) and not os.path.isfile(download_target):
raise RuntimeError(f"{download_target} exists and is not a regular file")
if os.path.isfile(download_target):
with open(download_target, "rb") as f:
model_bytes = f.read()
if hashlib.sha256(model_bytes).hexdigest() == expected_sha256:
return model_bytes if in_memory else download_target
else:
warnings.warn(f"{download_target} exists, but the SHA256 checksum does not match; re-downloading the file")
with urllib.request.urlopen(url) as source, open(download_target, "wb") as output:
with tqdm(total=int(source.info().get("Content-Length")), ncols=80, unit='iB', unit_scale=True, unit_divisor=1024) as loop:
while True:
buffer = source.read(8192)
if not buffer:
break
output.write(buffer)
loop.update(len(buffer))
model_bytes = open(download_target, "rb").read()
if hashlib.sha256(model_bytes).hexdigest() != expected_sha256:
raise RuntimeError("Model has been downloaded but the SHA256 checksum does not not match. Please retry loading the model.")
return model_bytes if in_memory else download_target
def available_models() -> List[str]:
"""Returns the names of available models"""
return list(_MODELS.keys())
def load_model(name: str, device: Optional[Union[str, torch.device]] = None, download_root: str = None, in_memory: bool = False) -> Whisper:
"""
Load a Whisper ASR model
Parameters
----------
name : str
one of the official model names listed by `whisper.available_models()`, or
path to a model checkpoint containing the model dimensions and the model state_dict.
device : Union[str, torch.device]
the PyTorch device to put the model into
download_root: str
path to download the model files; by default, it uses "~/.cache/whisper"
in_memory: bool
whether to preload the model weights into host memory
Returns
-------
model : Whisper
The Whisper ASR model instance
"""
if device is None:
device = "cuda" if torch.cuda.is_available() else "cpu"
if download_root is None:
download_root = os.getenv(
"XDG_CACHE_HOME",
os.path.join(os.path.expanduser("~"), ".cache", "whisper")
)
if name in _MODELS:
checkpoint_file = _download(_MODELS[name], download_root, in_memory)
elif os.path.isfile(name):
checkpoint_file = open(name, "rb").read() if in_memory else name
else:
raise RuntimeError(f"Model {name} not found; available models = {available_models()}")
with (io.BytesIO(checkpoint_file) if in_memory else open(checkpoint_file, "rb")) as fp:
checkpoint = torch.load(fp, map_location=device)
del checkpoint_file
dims = ModelDimensions(**checkpoint["dims"])
model = Whisper(dims)
model.load_state_dict(checkpoint["model_state_dict"])
return model.to(device)
from .vad import load_vad_model

2
whisperx/alignment.py
View File

@ -9,7 +9,7 @@ from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import torchaudio
import torch
from dataclasses import dataclass
from .audio import SAMPLE_RATE, load_audio
from whisper.audio import SAMPLE_RATE, load_audio
from .utils import interpolate_nans

428
whisperx/asr.py Normal file
View File

@ -0,0 +1,428 @@
import warnings
from typing import TYPE_CHECKING, Optional, Tuple, Union
import tempfile
import numpy as np
import torch
import tqdm
from whisper.audio import (
FRAMES_PER_SECOND,
HOP_LENGTH,
N_FRAMES,
N_SAMPLES,
SAMPLE_RATE,
CHUNK_LENGTH,
log_mel_spectrogram,
pad_or_trim,
load_audio
)
from whisper.decoding import DecodingOptions, DecodingResult
from whisper.timing import add_word_timestamps
from whisper.tokenizer import LANGUAGES, TO_LANGUAGE_CODE, get_tokenizer
from whisper.utils import (
exact_div,
format_timestamp,
make_safe,
)
if TYPE_CHECKING:
from whisper.model import Whisper
from .vad import merge_chunks
def transcribe(
model: "Whisper",
audio: Union[str, np.ndarray, torch.Tensor] = None,
mel: np.ndarray = None,
verbose: Optional[bool] = None,
temperature: Union[float, Tuple[float, ...]] = (0.0, 0.2, 0.4, 0.6, 0.8, 1.0),
compression_ratio_threshold: Optional[float] = 2.4,
logprob_threshold: Optional[float] = -1.0,
no_speech_threshold: Optional[float] = 0.6,
condition_on_previous_text: bool = True,
initial_prompt: Optional[str] = None,
word_timestamps: bool = False,
prepend_punctuations: str = "\"'“¿([{-",
append_punctuations: str = "\"'.。,，!！?？:：”)]}、",
**decode_options,
):
"""
Transcribe an audio file using Whisper.
We redefine the Whisper transcribe function to allow mel input (for sequential slicing of audio)
Parameters
----------
model: Whisper
The Whisper model instance
audio: Union[str, np.ndarray, torch.Tensor]
The path to the audio file to open, or the audio waveform
mel: np.ndarray
Mel spectrogram of audio segment.
verbose: bool
Whether to display the text being decoded to the console. If True, displays all the details,
If False, displays minimal details. If None, does not display anything
temperature: Union[float, Tuple[float, ...]]
Temperature for sampling. It can be a tuple of temperatures, which will be successively used
upon failures according to either `compression_ratio_threshold` or `logprob_threshold`.
compression_ratio_threshold: float
If the gzip compression ratio is above this value, treat as failed
logprob_threshold: float
If the average log probability over sampled tokens is below this value, treat as failed
no_speech_threshold: float
If the no_speech probability is higher than this value AND the average log probability
over sampled tokens is below `logprob_threshold`, consider the segment as silent
condition_on_previous_text: bool
if True, the previous output of the model is provided as a prompt for the next window;
disabling may make the text inconsistent across windows, but the model becomes less prone to
getting stuck in a failure loop, such as repetition looping or timestamps going out of sync.
word_timestamps: bool
Extract word-level timestamps using the cross-attention pattern and dynamic time warping,
and include the timestamps for each word in each segment.
prepend_punctuations: str
If word_timestamps is True, merge these punctuation symbols with the next word
append_punctuations: str
If word_timestamps is True, merge these punctuation symbols with the previous word
initial_prompt: Optional[str]
Optional text to provide as a prompt for the first window. This can be used to provide, or
"prompt-engineer" a context for transcription, e.g. custom vocabularies or proper nouns
to make it more likely to predict those word correctly.
decode_options: dict
Keyword arguments to construct `DecodingOptions` instances
Returns
-------
A dictionary containing the resulting text ("text") and segment-level details ("segments"), and
the spoken language ("language"), which is detected when `decode_options["language"]` is None.
"""
dtype = torch.float16 if decode_options.get("fp16", True) else torch.float32
if model.device == torch.device("cpu"):
if torch.cuda.is_available():
warnings.warn("Performing inference on CPU when CUDA is available")
if dtype == torch.float16:
warnings.warn("FP16 is not supported on CPU; using FP32 instead")
dtype = torch.float32
if dtype == torch.float32:
decode_options["fp16"] = False
# Pad 30-seconds of silence to the input audio, for slicing
if mel is None:
if audio is None:
raise ValueError("Transcribe needs either audio or mel as input, currently both are none.")
mel = log_mel_spectrogram(audio, padding=N_SAMPLES)
content_frames = mel.shape[-1] - N_FRAMES
if decode_options.get("language", None) is None:
if not model.is_multilingual:
decode_options["language"] = "en"
else:
if verbose:
print(
"Detecting language using up to the first 30 seconds. Use `--language` to specify the language"
)
mel_segment = pad_or_trim(mel, N_FRAMES).to(model.device).to(dtype)
_, probs = model.detect_language(mel_segment)
decode_options["language"] = max(probs, key=probs.get)
if verbose is not None:
print(
f"Detected language: {LANGUAGES[decode_options['language']].title()}"
)
language: str = decode_options["language"]
task: str = decode_options.get("task", "transcribe")
tokenizer = get_tokenizer(model.is_multilingual, language=language, task=task)
if word_timestamps and task == "translate":
warnings.warn("Word-level timestamps on translations may not be reliable.")
def decode_with_fallback(segment: torch.Tensor) -> DecodingResult:
temperatures = (
[temperature] if isinstance(temperature, (int, float)) else temperature
)
decode_result = None
for t in temperatures:
kwargs = {**decode_options}
if t > 0:
# disable beam_size and patience when t > 0
kwargs.pop("beam_size", None)
kwargs.pop("patience", None)
else:
# disable best_of when t == 0
kwargs.pop("best_of", None)
options = DecodingOptions(**kwargs, temperature=t)
decode_result = model.decode(segment, options)
needs_fallback = False
if (
compression_ratio_threshold is not None
and decode_result.compression_ratio > compression_ratio_threshold
):
needs_fallback = True # too repetitive
if (
logprob_threshold is not None
and decode_result.avg_logprob < logprob_threshold
):
needs_fallback = True # average log probability is too low
if not needs_fallback:
break
return decode_result
seek = 0
input_stride = exact_div(
N_FRAMES, model.dims.n_audio_ctx
) # mel frames per output token: 2
time_precision = (
input_stride * HOP_LENGTH / SAMPLE_RATE
) # time per output token: 0.02 (seconds)
all_tokens = []
all_segments = []
prompt_reset_since = 0
if initial_prompt is not None:
initial_prompt_tokens = tokenizer.encode(" " + initial_prompt.strip())
all_tokens.extend(initial_prompt_tokens)
else:
initial_prompt_tokens = []
def new_segment(
*, start: float, end: float, tokens: torch.Tensor, result: DecodingResult
):
tokens = tokens.tolist()
text_tokens = [token for token in tokens if token < tokenizer.eot]
return {
"seek": seek,
"start": start,
"end": end,
"text": tokenizer.decode(text_tokens),
"tokens": tokens,
"temperature": result.temperature,
"avg_logprob": result.avg_logprob,
"compression_ratio": result.compression_ratio,
"no_speech_prob": result.no_speech_prob,
}
# show the progress bar when verbose is False (if True, transcribed text will be printed)
with tqdm.tqdm(
total=content_frames, unit="frames", disable=verbose is not False
) as pbar:
while seek < content_frames:
time_offset = float(seek * HOP_LENGTH / SAMPLE_RATE)
mel_segment = mel[:, seek : seek + N_FRAMES]
segment_size = min(N_FRAMES, content_frames - seek)
segment_duration = segment_size * HOP_LENGTH / SAMPLE_RATE
mel_segment = pad_or_trim(mel_segment, N_FRAMES).to(model.device).to(dtype)
decode_options["prompt"] = all_tokens[prompt_reset_since:]
result: DecodingResult = decode_with_fallback(mel_segment)
tokens = torch.tensor(result.tokens)
if no_speech_threshold is not None:
# no voice activity check
should_skip = result.no_speech_prob > no_speech_threshold
if (
logprob_threshold is not None
and result.avg_logprob > logprob_threshold
):
# don't skip if the logprob is high enough, despite the no_speech_prob
should_skip = False
if should_skip:
seek += segment_size # fast-forward to the next segment boundary
continue
previous_seek = seek
current_segments = []
timestamp_tokens: torch.Tensor = tokens.ge(tokenizer.timestamp_begin)
single_timestamp_ending = timestamp_tokens[-2:].tolist() == [False, True]
consecutive = torch.where(timestamp_tokens[:-1] & timestamp_tokens[1:])[0]
consecutive.add_(1)
if len(consecutive) > 0:
# if the output contains two consecutive timestamp tokens
slices = consecutive.tolist()
if single_timestamp_ending:
slices.append(len(tokens))
last_slice = 0
for current_slice in slices:
sliced_tokens = tokens[last_slice:current_slice]
start_timestamp_pos = (
sliced_tokens[0].item() - tokenizer.timestamp_begin
)
end_timestamp_pos = (
sliced_tokens[-1].item() - tokenizer.timestamp_begin
)
current_segments.append(
new_segment(
start=time_offset + start_timestamp_pos * time_precision,
end=time_offset + end_timestamp_pos * time_precision,
tokens=sliced_tokens,
result=result,
)
)
last_slice = current_slice
if single_timestamp_ending:
# single timestamp at the end means no speech after the last timestamp.
seek += segment_size
else:
# otherwise, ignore the unfinished segment and seek to the last timestamp
last_timestamp_pos = (
tokens[last_slice - 1].item() - tokenizer.timestamp_begin
)
seek += last_timestamp_pos * input_stride
else:
duration = segment_duration
timestamps = tokens[timestamp_tokens.nonzero().flatten()]
if (
len(timestamps) > 0
and timestamps[-1].item() != tokenizer.timestamp_begin
):
# no consecutive timestamps but it has a timestamp; use the last one.
last_timestamp_pos = (
timestamps[-1].item() - tokenizer.timestamp_begin
)
duration = last_timestamp_pos * time_precision
current_segments.append(
new_segment(
start=time_offset,
end=time_offset + duration,
tokens=tokens,
result=result,
)
)
seek += segment_size
if not condition_on_previous_text or result.temperature > 0.5:
# do not feed the prompt tokens if a high temperature was used
prompt_reset_since = len(all_tokens)
if word_timestamps:
add_word_timestamps(
segments=current_segments,
model=model,
tokenizer=tokenizer,
mel=mel_segment,
num_frames=segment_size,
prepend_punctuations=prepend_punctuations,
append_punctuations=append_punctuations,
)
word_end_timestamps = [
w["end"] for s in current_segments for w in s["words"]
]
if not single_timestamp_ending and len(word_end_timestamps) > 0:
seek_shift = round(
(word_end_timestamps[-1] - time_offset) * FRAMES_PER_SECOND
)
if seek_shift > 0:
seek = previous_seek + seek_shift
if verbose:
for segment in current_segments:
start, end, text = segment["start"], segment["end"], segment["text"]
line = f"[{format_timestamp(start)} --> {format_timestamp(end)}] {text}"
print(make_safe(line))
# if a segment is instantaneous or does not contain text, clear it
for i, segment in enumerate(current_segments):
if segment["start"] == segment["end"] or segment["text"].strip() == "":
segment["text"] = ""
segment["tokens"] = []
segment["words"] = []
all_segments.extend(
[
{"id": i, **segment}
for i, segment in enumerate(
current_segments, start=len(all_segments)
)
]
)
all_tokens.extend(
[token for segment in current_segments for token in segment["tokens"]]
)
# update progress bar
pbar.update(min(content_frames, seek) - previous_seek)
return dict(
text=tokenizer.decode(all_tokens[len(initial_prompt_tokens) :]),
segments=all_segments,
language=language,
)
def transcribe_with_vad(
model: "Whisper",
audio: str,
vad_pipeline,
mel = None,
verbose: Optional[bool] = None,
**kwargs
):
"""
Transcribe per VAD segment
"""
vad_segments = vad_pipeline(audio)
# if not torch.is_tensor(audio):
# if isinstance(audio, str):
audio = load_audio(audio)
audio = torch.from_numpy(audio)
prev = 0
output = {"segments": []}
# merge segments to approx 30s inputs to make whisper most appropraite
vad_segments = merge_chunks(vad_segments, chunk_size=CHUNK_LENGTH)
print("Performing transcription...")
for sdx, seg_t in enumerate(vad_segments):
if verbose:
print(f"~~ Transcribing VAD chunk: ({format_timestamp(seg_t['start'])} --> {format_timestamp(seg_t['end'])}) ~~")
seg_f_start, seg_f_end = int(seg_t["start"] * SAMPLE_RATE), int(seg_t["end"] * SAMPLE_RATE)
local_f_start, local_f_end = seg_f_start - prev, seg_f_end - prev
audio = audio[local_f_start:] # seek forward
seg_audio = audio[:local_f_end-local_f_start] # seek forward
prev = seg_f_start
local_mel = log_mel_spectrogram(seg_audio, padding=N_SAMPLES)
# need to pad
result = transcribe(model, audio, mel=local_mel, verbose=verbose, **kwargs)
seg_t["text"] = result["text"]
output["segments"].append(
{
"start": seg_t["start"],
"end": seg_t["end"],
"language": result["language"],
"text": result["text"],
"seg-text": [x["text"] for x in result["segments"]],
"seg-start": [x["start"] for x in result["segments"]],
"seg-end": [x["end"] for x in result["segments"]],
}
)
output["language"] = output["segments"][0]["language"]
return output

50001
whisperx/assets/gpt2/merges.txt
View File

File diff suppressed because it is too large Load Diff

1
whisperx/assets/gpt2/special_tokens_map.json
View File

@ -1 +0,0 @@
{"bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "unk_token": "<|endoftext|>"}

1
whisperx/assets/gpt2/tokenizer_config.json
View File

@ -1 +0,0 @@
{"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "gpt2", "tokenizer_class": "GPT2Tokenizer"}

1
whisperx/assets/gpt2/vocab.json
View File

File diff suppressed because one or more lines are too long

BIN
whisperx/assets/mel_filters.npz
View File

Binary file not shown.

1
whisperx/assets/multilingual/added_tokens.json
View File

@ -1 +0,0 @@
{"<|endoftext|>": 50257}

50000
whisperx/assets/multilingual/merges.txt
View File

File diff suppressed because it is too large Load Diff

1
whisperx/assets/multilingual/special_tokens_map.json
View File

@ -1 +0,0 @@
{"bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "unk_token": "<|endoftext|>"}

1
whisperx/assets/multilingual/tokenizer_config.json
View File

@ -1 +0,0 @@
{"unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "bos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "multilingual", "errors": "replace", "tokenizer_class": "GPT2Tokenizer"}

1
whisperx/assets/multilingual/vocab.json
View File

File diff suppressed because one or more lines are too long

124
whisperx/audio.py
View File

@ -1,124 +0,0 @@
import os
from functools import lru_cache
from typing import Union
import ffmpeg
import numpy as np
import torch
import torch.nn.functional as F
from .utils import exact_div
# hard-coded audio hyperparameters
SAMPLE_RATE = 16000
N_FFT = 400
N_MELS = 80
HOP_LENGTH = 160
CHUNK_LENGTH = 30
N_SAMPLES = CHUNK_LENGTH * SAMPLE_RATE # 480000: number of samples in a chunk
N_FRAMES = exact_div(N_SAMPLES, HOP_LENGTH) # 3000: number of frames in a mel spectrogram input
def load_audio(file: str, sr: int = SAMPLE_RATE):
"""
Open an audio file and read as mono waveform, resampling as necessary
Parameters
----------
file: str
The audio file to open
sr: int
The sample rate to resample the audio if necessary
Returns
-------
A NumPy array containing the audio waveform, in float32 dtype.
"""
try:
# This launches a subprocess to decode audio while down-mixing and resampling as necessary.
# Requires the ffmpeg CLI and `ffmpeg-python` package to be installed.
out, _ = (
ffmpeg.input(file, threads=0)
.output("-", format="s16le", acodec="pcm_s16le", ac=1, ar=sr)
.run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
)
except ffmpeg.Error as e:
raise RuntimeError(f"Failed to load audio: {e.stderr.decode()}") from e
return np.frombuffer(out, np.int16).flatten().astype(np.float32) / 32768.0
def pad_or_trim(array, length: int = N_SAMPLES, *, axis: int = -1):
"""
Pad or trim the audio array to N_SAMPLES, as expected by the encoder.
"""
if torch.is_tensor(array):
if array.shape[axis] > length:
array = array.index_select(dim=axis, index=torch.arange(length, device=array.device))
if array.shape[axis] < length:
pad_widths = [(0, 0)] * array.ndim
pad_widths[axis] = (0, length - array.shape[axis])
array = F.pad(array, [pad for sizes in pad_widths[::-1] for pad in sizes])
else:
if array.shape[axis] > length:
array = array.take(indices=range(length), axis=axis)
if array.shape[axis] < length:
pad_widths = [(0, 0)] * array.ndim
pad_widths[axis] = (0, length - array.shape[axis])
array = np.pad(array, pad_widths)
return array
@lru_cache(maxsize=None)
def mel_filters(device, n_mels: int = N_MELS) -> torch.Tensor:
"""
load the mel filterbank matrix for projecting STFT into a Mel spectrogram.
Allows decoupling librosa dependency; saved using:
np.savez_compressed(
"mel_filters.npz",
mel_80=librosa.filters.mel(sr=16000, n_fft=400, n_mels=80),
)
"""
assert n_mels == 80, f"Unsupported n_mels: {n_mels}"
with np.load(os.path.join(os.path.dirname(__file__), "assets", "mel_filters.npz")) as f:
return torch.from_numpy(f[f"mel_{n_mels}"]).to(device)
def log_mel_spectrogram(audio: Union[str, np.ndarray, torch.Tensor], n_mels: int = N_MELS):
"""
Compute the log-Mel spectrogram of
Parameters
----------
audio: Union[str, np.ndarray, torch.Tensor], shape = (*)
The path to audio or either a NumPy array or Tensor containing the audio waveform in 16 kHz
n_mels: int
The number of Mel-frequency filters, only 80 is supported
Returns
-------
torch.Tensor, shape = (80, n_frames)
A Tensor that contains the Mel spectrogram
"""
if not torch.is_tensor(audio):
if isinstance(audio, str):
audio = load_audio(audio)
audio = torch.from_numpy(audio)
window = torch.hann_window(N_FFT).to(audio.device)
stft = torch.stft(audio, N_FFT, HOP_LENGTH, window=window, return_complex=True)
magnitudes = stft[..., :-1].abs() ** 2
filters = mel_filters(audio.device, n_mels)
mel_spec = filters @ magnitudes
log_spec = torch.clamp(mel_spec, min=1e-10).log10()
log_spec = torch.maximum(log_spec, log_spec.max() - 8.0)
log_spec = (log_spec + 4.0) / 4.0
return log_spec

716
whisperx/decoding.py
View File

@ -1,716 +0,0 @@
from dataclasses import dataclass, field
from typing import Dict, List, Tuple, Iterable, Optional, Sequence, Union, TYPE_CHECKING
import numpy as np
import torch
import torch.nn.functional as F
from torch import Tensor
from torch.distributions import Categorical
from .audio import CHUNK_LENGTH
from .tokenizer import Tokenizer, get_tokenizer
from .utils import compression_ratio
if TYPE_CHECKING:
from .model import Whisper
@torch.no_grad()
def detect_language(model: "Whisper", mel: Tensor, tokenizer: Tokenizer = None) -> Tuple[Tensor, List[dict]]:
"""
Detect the spoken language in the audio, and return them as list of strings, along with the ids
of the most probable language tokens and the probability distribution over all language tokens.
This is performed outside the main decode loop in order to not interfere with kv-caching.
Returns
-------
language_tokens : Tensor, shape = (n_audio,)
ids of the most probable language tokens, which appears after the startoftranscript token.
language_probs : List[Dict[str, float]], length = n_audio
list of dictionaries containing the probability distribution over all languages.
"""
if tokenizer is None:
tokenizer = get_tokenizer(model.is_multilingual)
if tokenizer.language is None or tokenizer.language_token not in tokenizer.sot_sequence:
raise ValueError(f"This model doesn't have language tokens so it can't perform lang id")
single = mel.ndim == 2
if single:
mel = mel.unsqueeze(0)
# skip encoder forward pass if already-encoded audio features were given
if mel.shape[-2:] != (model.dims.n_audio_ctx, model.dims.n_audio_state):
mel = model.encoder(mel)
# forward pass using a single token, startoftranscript
n_audio = mel.shape[0]
x = torch.tensor([[tokenizer.sot]] * n_audio).to(mel.device) # [n_audio, 1]
logits = model.logits(x, mel)[:, 0]
# collect detected languages; suppress all non-language tokens
mask = torch.ones(logits.shape[-1], dtype=torch.bool)
mask[list(tokenizer.all_language_tokens)] = False
logits[:, mask] = -np.inf
language_tokens = logits.argmax(dim=-1)
language_token_probs = logits.softmax(dim=-1).cpu()
language_probs = [
{
c: language_token_probs[i, j].item()
for j, c in zip(tokenizer.all_language_tokens, tokenizer.all_language_codes)
}
for i in range(n_audio)
]
if single:
language_tokens = language_tokens[0]
language_probs = language_probs[0]
return language_tokens, language_probs
@dataclass(frozen=True)
class DecodingOptions:
task: str = "transcribe" # whether to perform X->X "transcribe" or X->English "translate"
language: Optional[str] = None # language that the audio is in; uses detected language if None
# sampling-related options
temperature: float = 0.0
sample_len: Optional[int] = None # maximum number of tokens to sample
best_of: Optional[int] = None # number of independent samples to collect, when t > 0
beam_size: Optional[int] = None # number of beams in beam search, when t == 0
patience: Optional[float] = None # patience in beam search (https://arxiv.org/abs/2204.05424)
# options for ranking generations (either beams or best-of-N samples)
length_penalty: Optional[float] = None # "alpha" in Google NMT, None defaults to length norm
# prompt, prefix, and token suppression
prompt: Optional[Union[str, List[int]]] = None # text or tokens for the previous context
prefix: Optional[Union[str, List[int]]] = None # text or tokens to prefix the current context
suppress_blank: bool = True # this will suppress blank outputs
# list of tokens ids (or comma-separated token ids) to suppress
# "-1" will suppress a set of symbols as defined in `tokenizer.non_speech_tokens()`
suppress_tokens: Optional[Union[str, Iterable[int]]] = "-1"
# timestamp sampling options
without_timestamps: bool = False # use <|notimestamps|> to sample text tokens only
max_initial_timestamp: Optional[float] = 1.0 # the initial timestamp cannot be later than this
# implementation details
fp16: bool = True # use fp16 for most of the calculation
@dataclass(frozen=True)
class DecodingResult:
audio_features: Tensor
language: str
language_probs: Optional[Dict[str, float]] = None
tokens: List[int] = field(default_factory=list)
text: str = ""
avg_logprob: float = np.nan
no_speech_prob: float = np.nan
temperature: float = np.nan
compression_ratio: float = np.nan
class Inference:
def logits(self, tokens: Tensor, audio_features: Tensor) -> Tensor:
"""Perform a forward pass on the decoder and return per-token logits"""
raise NotImplementedError
def rearrange_kv_cache(self, source_indices) -> None:
"""Update the key-value cache according to the updated beams"""
raise NotImplementedError
def cleanup_caching(self) -> None:
"""Clean up any resources or hooks after decoding is finished"""
pass
class PyTorchInference(Inference):
def __init__(self, model: "Whisper", initial_token_length: int):
self.model: "Whisper" = model
self.initial_token_length = initial_token_length
self.kv_cache = {}
self.hooks = []
def logits(self, tokens: Tensor, audio_features: Tensor) -> Tensor:
if not self.kv_cache:
self.kv_cache, self.hooks = self.model.install_kv_cache_hooks()
if tokens.shape[-1] > self.initial_token_length:
# only need to use the last token except in the first forward pass
tokens = tokens[:, -1:]
return self.model.decoder(tokens, audio_features, kv_cache=self.kv_cache)
def cleanup_caching(self):
for hook in self.hooks:
hook.remove()
self.kv_cache = {}
self.hooks = []
def rearrange_kv_cache(self, source_indices):
for module, tensor in self.kv_cache.items():
# update the key/value cache to contain the selected sequences
self.kv_cache[module] = tensor[source_indices].detach()
class SequenceRanker:
def rank(self, tokens: List[List[Tensor]], sum_logprobs: List[List[float]]) -> List[int]:
"""
Given a list of groups of samples and their cumulative log probabilities,
return the indices of the samples in each group to select as the final result
"""
raise NotImplementedError
class MaximumLikelihoodRanker(SequenceRanker):
"""
Select the sample with the highest log probabilities, penalized using either
a simple length normalization or Google NMT paper's length penalty
"""
def __init__(self, length_penalty: Optional[float]):
self.length_penalty = length_penalty
def rank(self, tokens: List[List[Tensor]], sum_logprobs: List[List[float]]):
def scores(logprobs, lengths):
result = []
for logprob, length in zip(logprobs, lengths):
if self.length_penalty is None:
penalty = length
else:
# from the Google NMT paper
penalty = ((5 + length) / 6) ** self.length_penalty
result.append(logprob / penalty)
return result
# get the sequence with the highest score
lengths = [[len(t) for t in s] for s in tokens]
return [np.argmax(scores(p, l)) for p, l in zip(sum_logprobs, lengths)]
class TokenDecoder:
def reset(self):
"""Initialize any stateful variables for decoding a new sequence"""
def update(self, tokens: Tensor, logits: Tensor, sum_logprobs: Tensor) -> Tuple[Tensor, bool]:
"""Specify how to select the next token, based on the current trace and logits
Parameters
----------
tokens : Tensor, shape = (n_batch, current_sequence_length)
all tokens in the context so far, including the prefix and sot_sequence tokens
logits : Tensor, shape = (n_batch, vocab_size)
per-token logits of the probability distribution at the current step
sum_logprobs : Tensor, shape = (n_batch)
cumulative log probabilities for each sequence
Returns
-------
tokens : Tensor, shape = (n_batch, current_sequence_length + 1)
the tokens, appended with the selected next token
completed : bool
True if all sequences has reached the end of text
"""
raise NotImplementedError
def finalize(
self, tokens: Tensor, sum_logprobs: Tensor
) -> Tuple[Sequence[Sequence[Tensor]], List[List[float]]]:
"""Finalize search and return the final candidate sequences
Parameters
----------
tokens : Tensor, shape = (n_audio, n_group, current_sequence_length)
all tokens in the context so far, including the prefix and sot_sequence
sum_logprobs : Tensor, shape = (n_audio, n_group)
cumulative log probabilities for each sequence
Returns
-------
tokens : Sequence[Sequence[Tensor]], length = n_audio
sequence of Tensors containing candidate token sequences, for each audio input
sum_logprobs : List[List[float]], length = n_audio
sequence of cumulative log probabilities corresponding to the above
"""
raise NotImplementedError
class GreedyDecoder(TokenDecoder):
def __init__(self, temperature: float, eot: int):
self.temperature = temperature
self.eot = eot
def update(self, tokens: Tensor, logits: Tensor, sum_logprobs: Tensor) -> Tuple[Tensor, bool]:
temperature = self.temperature
if temperature == 0:
next_tokens = logits.argmax(dim=-1)
else:
next_tokens = Categorical(logits=logits / temperature).sample()
logprobs = F.log_softmax(logits.float(), dim=-1)
current_logprobs = logprobs[torch.arange(logprobs.shape[0]), next_tokens]
sum_logprobs += current_logprobs * (tokens[:, -1] != self.eot)
next_tokens[tokens[:, -1] == self.eot] = self.eot
tokens = torch.cat([tokens, next_tokens[:, None]], dim=-1)
completed = (tokens[:, -1] == self.eot).all()
return tokens, completed
def finalize(self, tokens: Tensor, sum_logprobs: Tensor):
# make sure each sequence has at least one EOT token at the end
tokens = F.pad(tokens, (0, 1), value=self.eot)
return tokens, sum_logprobs.tolist()
class BeamSearchDecoder(TokenDecoder):
def __init__(self, beam_size: int, eot: int, inference: Inference, patience: Optional[float] = None):
self.beam_size = beam_size
self.eot = eot
self.inference = inference
self.patience = patience or 1.0
self.max_candidates: int = round(beam_size * self.patience)
self.finished_sequences = None
assert self.max_candidates > 0, f"Invalid beam size ({beam_size}) or patience ({patience})"
def reset(self):
self.finished_sequences = None
def update(self, tokens: Tensor, logits: Tensor, sum_logprobs: Tensor) -> Tuple[Tensor, bool]:
if tokens.shape[0] % self.beam_size != 0:
raise ValueError(f"{tokens.shape}[0] % {self.beam_size} != 0")
n_audio = tokens.shape[0] // self.beam_size
if self.finished_sequences is None: # for the first update
self.finished_sequences = [{} for _ in range(n_audio)]
logprobs = F.log_softmax(logits.float(), dim=-1)
next_tokens, source_indices, finished_sequences = [], [], []
for i in range(n_audio):
scores, sources, finished = {}, {}, {}
# STEP 1: calculate the cumulative log probabilities for possible candidates
for j in range(self.beam_size):
idx = i * self.beam_size + j
prefix = tokens[idx].tolist()
for logprob, token in zip(*logprobs[idx].topk(self.beam_size + 1)):
new_logprob = (sum_logprobs[idx] + logprob).item()
sequence = tuple(prefix + [token.item()])
scores[sequence] = new_logprob
sources[sequence] = idx
# STEP 2: rank the candidates and keep the top beam_size sequences for each audio
saved = 0
for sequence in sorted(scores, key=scores.get, reverse=True):
if sequence[-1] == self.eot:
finished[sequence] = scores[sequence]
else:
sum_logprobs[len(next_tokens)] = scores[sequence]
next_tokens.append(sequence)
source_indices.append(sources[sequence])
saved += 1
if saved == self.beam_size:
break
finished_sequences.append(finished)
tokens = torch.tensor(next_tokens, device=tokens.device)
self.inference.rearrange_kv_cache(source_indices)
# add newly finished sequences to self.finished_sequences
assert len(self.finished_sequences) == len(finished_sequences)
for previously_finished, newly_finished in zip(self.finished_sequences, finished_sequences):
for seq in sorted(newly_finished, key=newly_finished.get, reverse=True):
if len(previously_finished) >= self.max_candidates:
break # the candidate list is full
previously_finished[seq] = newly_finished[seq]
# mark as completed if all audio has enough number of samples
completed = all(
len(sequences) >= self.max_candidates for sequences in self.finished_sequences
)
return tokens, completed
def finalize(self, preceding_tokens: Tensor, sum_logprobs: Tensor):
# collect all finished sequences, including patience, and add unfinished ones if not enough
sum_logprobs = sum_logprobs.cpu()
for i, sequences in enumerate(self.finished_sequences):
if len(sequences) < self.beam_size: # when not enough sequences are finished
for j in list(np.argsort(sum_logprobs[i]))[::-1]:
sequence = preceding_tokens[i, j].tolist() + [self.eot]
sequences[tuple(sequence)] = sum_logprobs[i][j].item()
if len(sequences) >= self.beam_size:
break
tokens: List[List[Tensor]] = [
[torch.tensor(seq) for seq in sequences.keys()] for sequences in self.finished_sequences
]
sum_logprobs: List[List[float]] = [
list(sequences.values()) for sequences in self.finished_sequences
]
return tokens, sum_logprobs
class LogitFilter:
def apply(self, logits: Tensor, tokens: Tensor) -> None:
"""Apply any filtering or masking to logits in-place
Parameters
----------
logits : Tensor, shape = (n_batch, vocab_size)
per-token logits of the probability distribution at the current step
tokens : Tensor, shape = (n_batch, current_sequence_length)
all tokens in the context so far, including the prefix and sot_sequence tokens
"""
raise NotImplementedError
class SuppressBlank(LogitFilter):
def __init__(self, tokenizer: Tokenizer, sample_begin: int):
self.tokenizer = tokenizer
self.sample_begin = sample_begin
def apply(self, logits: Tensor, tokens: Tensor):
if tokens.shape[1] == self.sample_begin:
logits[:, self.tokenizer.encode(" ") + [self.tokenizer.eot]] = -np.inf
class SuppressTokens(LogitFilter):
def __init__(self, suppress_tokens: Sequence[int]):
self.suppress_tokens = list(suppress_tokens)
def apply(self, logits: Tensor, tokens: Tensor):
logits[:, self.suppress_tokens] = -np.inf
class ApplyTimestampRules(LogitFilter):
def __init__(
self, tokenizer: Tokenizer, sample_begin: int, max_initial_timestamp_index: Optional[int]
):
self.tokenizer = tokenizer
self.sample_begin = sample_begin
self.max_initial_timestamp_index = max_initial_timestamp_index
def apply(self, logits: Tensor, tokens: Tensor):
# suppress <|notimestamps|> which is handled by without_timestamps
if self.tokenizer.no_timestamps is not None:
logits[:, self.tokenizer.no_timestamps] = -np.inf
# timestamps have to appear in pairs, except directly before EOT; mask logits accordingly
for k in range(tokens.shape[0]):
sampled_tokens = tokens[k, self.sample_begin :]
seq = [t for t in sampled_tokens.tolist()]
last_was_timestamp = len(seq) >= 1 and seq[-1] >= self.tokenizer.timestamp_begin
penultimate_was_timestamp = len(seq) < 2 or seq[-2] >= self.tokenizer.timestamp_begin
if last_was_timestamp:
if penultimate_was_timestamp: # has to be non-timestamp
logits[k, self.tokenizer.timestamp_begin :] = -np.inf
else: # cannot be normal text tokens
logits[k, : self.tokenizer.eot] = -np.inf
timestamps = sampled_tokens[sampled_tokens.ge(self.tokenizer.timestamp_begin)]
if timestamps.numel() > 0:
# timestamps shouldn't decrease; forbid timestamp tokens smaller than the last
logits[k, self.tokenizer.timestamp_begin : timestamps[-1]] = -np.inf
if tokens.shape[1] == self.sample_begin:
# suppress generating non-timestamp tokens at the beginning
logits[:, : self.tokenizer.timestamp_begin] = -np.inf
# apply the `max_initial_timestamp` option
if self.max_initial_timestamp_index is not None:
last_allowed = self.tokenizer.timestamp_begin + self.max_initial_timestamp_index
logits[:, last_allowed + 1 :] = -np.inf
# if sum of probability over timestamps is above any other token, sample timestamp
logprobs = F.log_softmax(logits.float(), dim=-1)
for k in range(tokens.shape[0]):
timestamp_logprob = logprobs[k, self.tokenizer.timestamp_begin :].logsumexp(dim=-1)
max_text_token_logprob = logprobs[k, : self.tokenizer.timestamp_begin].max()
if timestamp_logprob > max_text_token_logprob:
logits[k, : self.tokenizer.timestamp_begin] = -np.inf
class DecodingTask:
inference: Inference
sequence_ranker: SequenceRanker
decoder: TokenDecoder
logit_filters: List[LogitFilter]
def __init__(self, model: "Whisper", options: DecodingOptions):
self.model = model
language = options.language or "en"
tokenizer = get_tokenizer(model.is_multilingual, language=language, task=options.task)
self.tokenizer: Tokenizer = tokenizer
self.options: DecodingOptions = self._verify_options(options)
self.n_group: int = options.beam_size or options.best_of or 1
self.n_ctx: int = model.dims.n_text_ctx
self.sample_len: int = options.sample_len or model.dims.n_text_ctx // 2
self.sot_sequence: Tuple[int] = tokenizer.sot_sequence
if self.options.without_timestamps:
self.sot_sequence = tokenizer.sot_sequence_including_notimestamps
self.initial_tokens: Tuple[int] = self._get_initial_tokens()
self.sample_begin: int = len(self.initial_tokens)
self.sot_index: int = self.initial_tokens.index(tokenizer.sot)
# inference: implements the forward pass through the decoder, including kv caching
self.inference = PyTorchInference(model, len(self.initial_tokens))
# sequence ranker: implements how to rank a group of sampled sequences
self.sequence_ranker = MaximumLikelihoodRanker(options.length_penalty)
# decoder: implements how to select the next tokens, given the autoregressive distribution
if options.beam_size is not None:
self.decoder = BeamSearchDecoder(
options.beam_size, tokenizer.eot, self.inference, options.patience
)
else:
self.decoder = GreedyDecoder(options.temperature, tokenizer.eot)
# logit filters: applies various rules to suppress or penalize certain tokens
self.logit_filters = []
if self.options.suppress_blank:
self.logit_filters.append(SuppressBlank(self.tokenizer, self.sample_begin))
if self.options.suppress_tokens:
self.logit_filters.append(SuppressTokens(self._get_suppress_tokens()))
if not options.without_timestamps:
precision = CHUNK_LENGTH / model.dims.n_audio_ctx # usually 0.02 seconds
max_initial_timestamp_index = None
if options.max_initial_timestamp:
max_initial_timestamp_index = round(self.options.max_initial_timestamp / precision)
self.logit_filters.append(
ApplyTimestampRules(tokenizer, self.sample_begin, max_initial_timestamp_index)
)
def _verify_options(self, options: DecodingOptions) -> DecodingOptions:
if options.beam_size is not None and options.best_of is not None:
raise ValueError("beam_size and best_of can't be given together")
if options.temperature == 0:
if options.best_of is not None:
raise ValueError("best_of with greedy sampling (T=0) is not compatible")
if options.patience is not None and options.beam_size is None:
raise ValueError("patience requires beam_size to be given")
if options.length_penalty is not None and not (0 <= options.length_penalty <= 1):
raise ValueError("length_penalty (alpha) should be a value between 0 and 1")
return options
def _get_initial_tokens(self) -> Tuple[int]:
tokens = list(self.sot_sequence)
prefix = self.options.prefix
prompt = self.options.prompt
if prefix:
prefix_tokens = (
self.tokenizer.encode(" " + prefix.strip()) if isinstance(prefix, str) else prefix
)
if self.sample_len is not None:
max_prefix_len = self.n_ctx // 2 - self.sample_len
prefix_tokens = prefix_tokens[-max_prefix_len:]
tokens = tokens + prefix_tokens
if prompt:
prompt_tokens = (
self.tokenizer.encode(" " + prompt.strip()) if isinstance(prompt, str) else prompt
)
tokens = [self.tokenizer.sot_prev] + prompt_tokens[-(self.n_ctx // 2 - 1) :] + tokens
return tuple(tokens)
def _get_suppress_tokens(self) -> Tuple[int]:
suppress_tokens = self.options.suppress_tokens
if isinstance(suppress_tokens, str):
suppress_tokens = [int(t) for t in suppress_tokens.split(",")]
if -1 in suppress_tokens:
suppress_tokens = [t for t in suppress_tokens if t >= 0]
suppress_tokens.extend(self.tokenizer.non_speech_tokens)
elif suppress_tokens is None or len(suppress_tokens) == 0:
suppress_tokens = [] # interpret empty string as an empty list
else:
assert isinstance(suppress_tokens, list), "suppress_tokens must be a list"
suppress_tokens.extend(
[self.tokenizer.sot, self.tokenizer.sot_prev, self.tokenizer.sot_lm]
)
if self.tokenizer.no_speech is not None:
# no-speech probability is collected separately
suppress_tokens.append(self.tokenizer.no_speech)
return tuple(sorted(set(suppress_tokens)))
def _get_audio_features(self, mel: Tensor):
if self.options.fp16:
mel = mel.half()
if mel.shape[-2:] == (self.model.dims.n_audio_ctx, self.model.dims.n_audio_state):
# encoded audio features are given; skip audio encoding
audio_features = mel
else:
audio_features = self.model.encoder(mel)
if audio_features.dtype != (torch.float16 if self.options.fp16 else torch.float32):
return TypeError(f"audio_features has an incorrect dtype: {audio_features.dtype}")
return audio_features
def _detect_language(self, audio_features: Tensor, tokens: Tensor):
languages = [self.options.language] * audio_features.shape[0]
lang_probs = None
if self.options.language is None or self.options.task == "lang_id":
lang_tokens, lang_probs = self.model.detect_language(audio_features, self.tokenizer)
languages = [max(probs, key=probs.get) for probs in lang_probs]
if self.options.language is None:
tokens[:, self.sot_index + 1] = lang_tokens # write language tokens
return languages, lang_probs
def _main_loop(self, audio_features: Tensor, tokens: Tensor):
assert audio_features.shape[0] == tokens.shape[0]
n_batch = tokens.shape[0]
sum_logprobs: Tensor = torch.zeros(n_batch, device=audio_features.device)
no_speech_probs = [np.nan] * n_batch
try:
for i in range(self.sample_len):
logits = self.inference.logits(tokens, audio_features)
if i == 0 and self.tokenizer.no_speech is not None: # save no_speech_probs
probs_at_sot = logits[:, self.sot_index].float().softmax(dim=-1)
no_speech_probs = probs_at_sot[:, self.tokenizer.no_speech].tolist()
# now we need to consider the logits at the last token only
logits = logits[:, -1]
# apply the logit filters, e.g. for suppressing or applying penalty to
for logit_filter in self.logit_filters:
logit_filter.apply(logits, tokens)
# expand the tokens tensor with the selected next tokens
tokens, completed = self.decoder.update(tokens, logits, sum_logprobs)
if completed or tokens.shape[-1] > self.n_ctx:
break
finally:
self.inference.cleanup_caching()
return tokens, sum_logprobs, no_speech_probs
@torch.no_grad()
def run(self, mel: Tensor) -> List[DecodingResult]:
self.decoder.reset()
tokenizer: Tokenizer = self.tokenizer
n_audio: int = mel.shape[0]
audio_features: Tensor = self._get_audio_features(mel) # encoder forward pass
tokens: Tensor = torch.tensor([self.initial_tokens]).repeat(n_audio, 1)
# detect language if requested, overwriting the language token
languages, language_probs = self._detect_language(audio_features, tokens)
if self.options.task == "lang_id":
return [
DecodingResult(audio_features=features, language=language, language_probs=probs)
for features, language, probs in zip(audio_features, languages, language_probs)
]
# repeat the audio & text tensors by the group size, for beam search or best-of-n sampling
audio_features = audio_features.repeat_interleave(self.n_group, dim=0)
tokens = tokens.repeat_interleave(self.n_group, dim=0).to(audio_features.device)
# call the main sampling loop
tokens, sum_logprobs, no_speech_probs = self._main_loop(audio_features, tokens)
# reshape the tensors to have (n_audio, n_group) as the first two dimensions
audio_features = audio_features[:: self.n_group]
no_speech_probs = no_speech_probs[:: self.n_group]
assert audio_features.shape[0] == len(no_speech_probs) == n_audio
tokens = tokens.reshape(n_audio, self.n_group, -1)
sum_logprobs = sum_logprobs.reshape(n_audio, self.n_group)
# get the final candidates for each group, and slice between the first sampled token and EOT
tokens, sum_logprobs = self.decoder.finalize(tokens, sum_logprobs)
tokens: List[List[Tensor]] = [
[t[self.sample_begin : (t == tokenizer.eot).nonzero()[0, 0]] for t in s] for s in tokens
]
# select the top-ranked sample in each group
selected = self.sequence_ranker.rank(tokens, sum_logprobs)
tokens: List[List[int]] = [t[i].tolist() for i, t in zip(selected, tokens)]
texts: List[str] = [tokenizer.decode(t).strip() for t in tokens]
sum_logprobs: List[float] = [lp[i] for i, lp in zip(selected, sum_logprobs)]
avg_logprobs: List[float] = [lp / (len(t) + 1) for t, lp in zip(tokens, sum_logprobs)]
fields = (texts, languages, tokens, audio_features, avg_logprobs, no_speech_probs)
if len(set(map(len, fields))) != 1:
raise RuntimeError(f"inconsistent result lengths: {list(map(len, fields))}")
return [
DecodingResult(
audio_features=features,
language=language,
tokens=tokens,
text=text,
avg_logprob=avg_logprob,
no_speech_prob=no_speech_prob,
temperature=self.options.temperature,
compression_ratio=compression_ratio(text),
)
for text, language, tokens, features, avg_logprob, no_speech_prob in zip(*fields)
]
@torch.no_grad()
def decode(model: "Whisper", mel: Tensor, options: DecodingOptions = DecodingOptions()) -> Union[DecodingResult, List[DecodingResult]]:
"""
Performs decoding of 30-second audio segment(s), provided as Mel spectrogram(s).
Parameters
----------
model: Whisper
the Whisper model instance
mel: torch.Tensor, shape = (80, 3000) or (*, 80, 3000)
A tensor containing the Mel spectrogram(s)
options: DecodingOptions
A dataclass that contains all necessary options for decoding 30-second segments
Returns
-------
result: Union[DecodingResult, List[DecodingResult]]
The result(s) of decoding contained in `DecodingResult` dataclass instance(s)
"""
single = mel.ndim == 2
if single:
mel = mel.unsqueeze(0)
result = DecodingTask(model, options).run(mel)
if single:
result = result[0]
return result

17
whisperx/diarize.py
View File

@ -1,8 +1,23 @@
import numpy as np
import pandas as pd
from pyannote.audio import Pipeline
class DiarizationPipeline:
def __init__(
self,
model_name="pyannote/speaker-diarization@2.1",
use_auth_token=None,
):
self.model = Pipeline.from_pretrained(model_name, use_auth_token=use_auth_token)
def __call__(self, audio, min_speakers=None, max_speakers=None):
segments = self.model(audio, min_speakers=min_speakers, max_speakers=max_speakers)
diarize_df = pd.DataFrame(segments.itertracks(yield_label=True))
diarize_df['start'] = diarize_df[0].apply(lambda x: x.start)
diarize_df['end'] = diarize_df[0].apply(lambda x: x.end)
return diarize_df
def assign_word_speakers(diarize_df, result_segments, fill_nearest=False):
for seg in result_segments:
wdf = seg['word-segments']
if len(wdf['start'].dropna()) == 0:

268
whisperx/model.py
View File

@ -1,268 +0,0 @@
from dataclasses import dataclass
from typing import Dict
from typing import Iterable, Optional
import numpy as np
import torch
import torch.nn.functional as F
from torch import Tensor
from torch import nn
from .transcribe import transcribe as transcribe_function
from .decoding import detect_language as detect_language_function, decode as decode_function
@dataclass
class ModelDimensions:
n_mels: int
n_audio_ctx: int
n_audio_state: int
n_audio_head: int
n_audio_layer: int
n_vocab: int
n_text_ctx: int
n_text_state: int
n_text_head: int
n_text_layer: int
class LayerNorm(nn.LayerNorm):
def forward(self, x: Tensor) -> Tensor:
return super().forward(x.float()).type(x.dtype)
class Linear(nn.Linear):
def forward(self, x: Tensor) -> Tensor:
return F.linear(
x, self.weight.to(x.dtype), None if self.bias is None else self.bias.to(x.dtype)
)
class Conv1d(nn.Conv1d):
def _conv_forward(self, x: Tensor, weight: Tensor, bias: Optional[Tensor]) -> Tensor:
return super()._conv_forward(
x, weight.to(x.dtype), None if bias is None else bias.to(x.dtype)
)
def sinusoids(length, channels, max_timescale=10000):
"""Returns sinusoids for positional embedding"""
assert channels % 2 == 0
log_timescale_increment = np.log(max_timescale) / (channels // 2 - 1)
inv_timescales = torch.exp(-log_timescale_increment * torch.arange(channels // 2))
scaled_time = torch.arange(length)[:, np.newaxis] * inv_timescales[np.newaxis, :]
return torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], dim=1)
class MultiHeadAttention(nn.Module):
def __init__(self, n_state: int, n_head: int):
super().__init__()
self.n_head = n_head
self.query = Linear(n_state, n_state)
self.key = Linear(n_state, n_state, bias=False)
self.value = Linear(n_state, n_state)
self.out = Linear(n_state, n_state)
def forward(
self,
x: Tensor,
xa: Optional[Tensor] = None,
mask: Optional[Tensor] = None,
kv_cache: Optional[dict] = None,
):
q = self.query(x)
if kv_cache is None or xa is None or self.key not in kv_cache:
# hooks, if installed (i.e. kv_cache is not None), will prepend the cached kv tensors;
# otherwise, perform key/value projections for self- or cross-attention as usual.
k = self.key(x if xa is None else xa)
v = self.value(x if xa is None else xa)
else:
# for cross-attention, calculate keys and values once and reuse in subsequent calls.
k = kv_cache[self.key]
v = kv_cache[self.value]
wv, qk = self.qkv_attention(q, k, v, mask)
return self.out(wv), qk
def qkv_attention(self, q: Tensor, k: Tensor, v: Tensor, mask: Optional[Tensor] = None):
n_batch, n_ctx, n_state = q.shape
scale = (n_state // self.n_head) ** -0.25
q = q.view(*q.shape[:2], self.n_head, -1).permute(0, 2, 1, 3) * scale
k = k.view(*k.shape[:2], self.n_head, -1).permute(0, 2, 3, 1) * scale
v = v.view(*v.shape[:2], self.n_head, -1).permute(0, 2, 1, 3)
qk = q @ k
if mask is not None:
qk = qk + mask[:n_ctx, :n_ctx]
qk = qk.float()
w = F.softmax(qk, dim=-1).to(q.dtype)
return (w @ v).permute(0, 2, 1, 3).flatten(start_dim=2), qk.detach()
class ResidualAttentionBlock(nn.Module):
def __init__(self, n_state: int, n_head: int, cross_attention: bool = False):
super().__init__()
self.attn = MultiHeadAttention(n_state, n_head)
self.attn_ln = LayerNorm(n_state)
self.cross_attn = MultiHeadAttention(n_state, n_head) if cross_attention else None
self.cross_attn_ln = LayerNorm(n_state) if cross_attention else None
n_mlp = n_state * 4
self.mlp = nn.Sequential(Linear(n_state, n_mlp), nn.GELU(), Linear(n_mlp, n_state))
self.mlp_ln = LayerNorm(n_state)
def forward(
self,
x: Tensor,
xa: Optional[Tensor] = None,
mask: Optional[Tensor] = None,
kv_cache: Optional[dict] = None,
):
x = x + self.attn(self.attn_ln(x), mask=mask, kv_cache=kv_cache)[0]
if self.cross_attn:
x = x + self.cross_attn(self.cross_attn_ln(x), xa, kv_cache=kv_cache)[0]
x = x + self.mlp(self.mlp_ln(x))
return x
class AudioEncoder(nn.Module):
def __init__(self, n_mels: int, n_ctx: int, n_state: int, n_head: int, n_layer: int):
super().__init__()
self.conv1 = Conv1d(n_mels, n_state, kernel_size=3, padding=1)
self.conv2 = Conv1d(n_state, n_state, kernel_size=3, stride=2, padding=1)
self.register_buffer("positional_embedding", sinusoids(n_ctx, n_state))
self.blocks: Iterable[ResidualAttentionBlock] = nn.ModuleList(
[ResidualAttentionBlock(n_state, n_head) for _ in range(n_layer)]
)
self.ln_post = LayerNorm(n_state)
def forward(self, x: Tensor):
"""
x : torch.Tensor, shape = (batch_size, n_mels, n_ctx)
the mel spectrogram of the audio
"""
x = F.gelu(self.conv1(x))
x = F.gelu(self.conv2(x))
x = x.permute(0, 2, 1)
assert x.shape[1:] == self.positional_embedding.shape, "incorrect audio shape"
x = (x + self.positional_embedding).to(x.dtype)
for block in self.blocks:
x = block(x)
x = self.ln_post(x)
return x
class TextDecoder(nn.Module):
def __init__(self, n_vocab: int, n_ctx: int, n_state: int, n_head: int, n_layer: int):
super().__init__()
self.token_embedding = nn.Embedding(n_vocab, n_state)
self.positional_embedding = nn.Parameter(torch.empty(n_ctx, n_state))
self.blocks: Iterable[ResidualAttentionBlock] = nn.ModuleList(
[ResidualAttentionBlock(n_state, n_head, cross_attention=True) for _ in range(n_layer)]
)
self.ln = LayerNorm(n_state)
mask = torch.empty(n_ctx, n_ctx).fill_(-np.inf).triu_(1)
self.register_buffer("mask", mask, persistent=False)
def forward(self, x: Tensor, xa: Tensor, kv_cache: Optional[dict] = None):
"""
x : torch.LongTensor, shape = (batch_size, <= n_ctx)
the text tokens
xa : torch.Tensor, shape = (batch_size, n_mels, n_audio_ctx)
the encoded audio features to be attended on
"""
offset = next(iter(kv_cache.values())).shape[1] if kv_cache else 0
x = self.token_embedding(x) + self.positional_embedding[offset : offset + x.shape[-1]]
x = x.to(xa.dtype)
for block in self.blocks:
x = block(x, xa, mask=self.mask, kv_cache=kv_cache)
x = self.ln(x)
logits = (x @ torch.transpose(self.token_embedding.weight.to(x.dtype), 0, 1)).float()
return logits
class Whisper(nn.Module):
def __init__(self, dims: ModelDimensions):
super().__init__()
self.dims = dims
self.encoder = AudioEncoder(
self.dims.n_mels,
self.dims.n_audio_ctx,
self.dims.n_audio_state,
self.dims.n_audio_head,
self.dims.n_audio_layer,
)
self.decoder = TextDecoder(
self.dims.n_vocab,
self.dims.n_text_ctx,
self.dims.n_text_state,
self.dims.n_text_head,
self.dims.n_text_layer,
)
def embed_audio(self, mel: torch.Tensor):
return self.encoder(mel)
def logits(self, tokens: torch.Tensor, audio_features: torch.Tensor):
return self.decoder(tokens, audio_features)
def forward(self, mel: torch.Tensor, tokens: torch.Tensor) -> Dict[str, torch.Tensor]:
return self.decoder(tokens, self.encoder(mel))
@property
def device(self):
return next(self.parameters()).device
@property
def is_multilingual(self):
return self.dims.n_vocab == 51865
def install_kv_cache_hooks(self, cache: Optional[dict] = None):
"""
The `MultiHeadAttention` module optionally accepts `kv_cache` which stores the key and value
tensors calculated for the previous positions. This method returns a dictionary that stores
all caches, and the necessary hooks for the key and value projection modules that save the
intermediate tensors to be reused during later calculations.
Returns
-------
cache : Dict[nn.Module, torch.Tensor]
A dictionary object mapping the key/value projection modules to its cache
hooks : List[RemovableHandle]
List of PyTorch RemovableHandle objects to stop the hooks to be called
"""
cache = {**cache} if cache is not None else {}
hooks = []
def save_to_cache(module, _, output):
if module not in cache or output.shape[1] > self.decoder.positional_embedding.shape[0]:
cache[module] = output # save as-is, for the first token or cross attention
else:
cache[module] = torch.cat([cache[module], output], dim=1).detach()
return cache[module]
def install_hooks(layer: nn.Module):
if isinstance(layer, MultiHeadAttention):
hooks.append(layer.key.register_forward_hook(save_to_cache))
hooks.append(layer.value.register_forward_hook(save_to_cache))
self.decoder.apply(install_hooks)
return cache, hooks
detect_language = detect_language_function
transcribe = transcribe_function
decode = decode_function

2
whisperx/normalizers/__init__.py
View File

@ -1,2 +0,0 @@
from .basic import BasicTextNormalizer
from .english import EnglishTextNormalizer

71
whisperx/normalizers/basic.py
View File

@ -1,71 +0,0 @@
import re
import unicodedata
import regex
# non-ASCII letters that are not separated by "NFKD" normalization
ADDITIONAL_DIACRITICS = {
"œ": "oe",
"Œ": "OE",
"ø": "o",
"Ø": "O",
"æ": "ae",
"Æ": "AE",
"ß": "ss",
"ẞ": "SS",
"đ": "d",
"Đ": "D",
"ð": "d",
"Ð": "D",
"þ": "th",
"Þ": "th",
"ł": "l",
"Ł": "L",
}
def remove_symbols_and_diacritics(s: str, keep=""):
"""
Replace any other markers, symbols, and punctuations with a space,
and drop any diacritics (category 'Mn' and some manual mappings)
"""
return "".join(
c
if c in keep
else ADDITIONAL_DIACRITICS[c]
if c in ADDITIONAL_DIACRITICS
else ""
if unicodedata.category(c) == "Mn"
else " "
if unicodedata.category(c)[0] in "MSP"
else c
for c in unicodedata.normalize("NFKD", s)
)
def remove_symbols(s: str):
"""
Replace any other markers, symbols, punctuations with a space, keeping diacritics
"""
return "".join(
" " if unicodedata.category(c)[0] in "MSP" else c for c in unicodedata.normalize("NFKC", s)
)
class BasicTextNormalizer:
def __init__(self, remove_diacritics: bool = False, split_letters: bool = False):
self.clean = remove_symbols_and_diacritics if remove_diacritics else remove_symbols
self.split_letters = split_letters
def __call__(self, s: str):
s = s.lower()
s = re.sub(r"[<\[][^>\]]*[>\]]", "", s) # remove words between brackets
s = re.sub(r"\(([^)]+?)\)", "", s) # remove words between parenthesis
s = self.clean(s).lower()
if self.split_letters:
s = " ".join(regex.findall(r"\X", s, regex.U))
s = re.sub(r"\s+", " ", s) # replace any successive whitespace characters with a space
return s

1741
whisperx/normalizers/english.json
View File

File diff suppressed because it is too large Load Diff

543
whisperx/normalizers/english.py
View File

@ -1,543 +0,0 @@
import json
import os
import re
from fractions import Fraction
from typing import Iterator, List, Match, Optional, Union
from more_itertools import windowed
from .basic import remove_symbols_and_diacritics
class EnglishNumberNormalizer:
"""
Convert any spelled-out numbers into arabic numbers, while handling:
- remove any commas
- keep the suffixes such as: `1960s`, `274th`, `32nd`, etc.
- spell out currency symbols after the number. e.g. `$20 million` -> `20000000 dollars`
- spell out `one` and `ones`
- interpret successive single-digit numbers as nominal: `one oh one` -> `101`
"""
def __init__(self):
super().__init__()
self.zeros = {"o", "oh", "zero"}
self.ones = {
name: i
for i, name in enumerate(
[
"one",
"two",
"three",
"four",
"five",
"six",
"seven",
"eight",
"nine",
"ten",
"eleven",
"twelve",
"thirteen",
"fourteen",
"fifteen",
"sixteen",
"seventeen",
"eighteen",
"nineteen",
],
start=1,
)
}
self.ones_plural = {
"sixes" if name == "six" else name + "s": (value, "s")
for name, value in self.ones.items()
}
self.ones_ordinal = {
"zeroth": (0, "th"),
"first": (1, "st"),
"second": (2, "nd"),
"third": (3, "rd"),
"fifth": (5, "th"),
"twelfth": (12, "th"),
**{
name + ("h" if name.endswith("t") else "th"): (value, "th")
for name, value in self.ones.items()
if value > 3 and value != 5 and value != 12
},
}
self.ones_suffixed = {**self.ones_plural, **self.ones_ordinal}
self.tens = {
"twenty": 20,
"thirty": 30,
"forty": 40,
"fifty": 50,
"sixty": 60,
"seventy": 70,
"eighty": 80,
"ninety": 90,
}
self.tens_plural = {
name.replace("y", "ies"): (value, "s") for name, value in self.tens.items()
}
self.tens_ordinal = {
name.replace("y", "ieth"): (value, "th") for name, value in self.tens.items()
}
self.tens_suffixed = {**self.tens_plural, **self.tens_ordinal}
self.multipliers = {
"hundred": 100,
"thousand": 1_000,
"million": 1_000_000,
"billion": 1_000_000_000,
"trillion": 1_000_000_000_000,
"quadrillion": 1_000_000_000_000_000,
"quintillion": 1_000_000_000_000_000_000,
"sextillion": 1_000_000_000_000_000_000_000,
"septillion": 1_000_000_000_000_000_000_000_000,
"octillion": 1_000_000_000_000_000_000_000_000_000,
"nonillion": 1_000_000_000_000_000_000_000_000_000_000,
"decillion": 1_000_000_000_000_000_000_000_000_000_000_000,
}
self.multipliers_plural = {
name + "s": (value, "s") for name, value in self.multipliers.items()
}
self.multipliers_ordinal = {
name + "th": (value, "th") for name, value in self.multipliers.items()
}
self.multipliers_suffixed = {**self.multipliers_plural, **self.multipliers_ordinal}
self.decimals = {*self.ones, *self.tens, *self.zeros}
self.preceding_prefixers = {
"minus": "-",
"negative": "-",
"plus": "+",
"positive": "+",
}
self.following_prefixers = {
"pound": "£",
"pounds": "£",
"euro": "€",
"euros": "€",
"dollar": "$",
"dollars": "$",
"cent": "¢",
"cents": "¢",
}
self.prefixes = set(
list(self.preceding_prefixers.values()) + list(self.following_prefixers.values())
)
self.suffixers = {
"per": {"cent": "%"},
"percent": "%",
}
self.specials = {"and", "double", "triple", "point"}
self.words = set(
[
key
for mapping in [
self.zeros,
self.ones,
self.ones_suffixed,
self.tens,
self.tens_suffixed,
self.multipliers,
self.multipliers_suffixed,
self.preceding_prefixers,
self.following_prefixers,
self.suffixers,
self.specials,
]
for key in mapping
]
)
self.literal_words = {"one", "ones"}
def process_words(self, words: List[str]) -> Iterator[str]:
prefix: Optional[str] = None
value: Optional[Union[str, int]] = None
skip = False
def to_fraction(s: str):
try:
return Fraction(s)
except ValueError:
return None
def output(result: Union[str, int]):
nonlocal prefix, value
result = str(result)
if prefix is not None:
result = prefix + result
value = None
prefix = None
return result
if len(words) == 0:
return
for prev, current, next in windowed([None] + words + [None], 3):
if skip:
skip = False
continue
next_is_numeric = next is not None and re.match(r"^\d+(\.\d+)?$", next)
has_prefix = current[0] in self.prefixes
current_without_prefix = current[1:] if has_prefix else current
if re.match(r"^\d+(\.\d+)?$", current_without_prefix):
# arabic numbers (potentially with signs and fractions)
f = to_fraction(current_without_prefix)
assert f is not None
if value is not None:
if isinstance(value, str) and value.endswith("."):
# concatenate decimals / ip address components
value = str(value) + str(current)
continue
else:
yield output(value)
prefix = current[0] if has_prefix else prefix
if f.denominator == 1:
value = f.numerator # store integers as int
else:
value = current_without_prefix
elif current not in self.words:
# non-numeric words
if value is not None:
yield output(value)
yield output(current)
elif current in self.zeros:
value = str(value or "") + "0"
elif current in self.ones:
ones = self.ones[current]
if value is None:
value = ones
elif isinstance(value, str) or prev in self.ones:
if prev in self.tens and ones < 10: # replace the last zero with the digit
assert value[-1] == "0"
value = value[:-1] + str(ones)
else:
value = str(value) + str(ones)
elif ones < 10:
if value % 10 == 0:
value += ones
else:
value = str(value) + str(ones)
else: # eleven to nineteen
if value % 100 == 0:
value += ones
else:
value = str(value) + str(ones)
elif current in self.ones_suffixed:
# ordinal or cardinal; yield the number right away
ones, suffix = self.ones_suffixed[current]
if value is None:
yield output(str(ones) + suffix)
elif isinstance(value, str) or prev in self.ones:
if prev in self.tens and ones < 10:
assert value[-1] == "0"
yield output(value[:-1] + str(ones) + suffix)
else:
yield output(str(value) + str(ones) + suffix)
elif ones < 10:
if value % 10 == 0:
yield output(str(value + ones) + suffix)
else:
yield output(str(value) + str(ones) + suffix)
else: # eleven to nineteen
if value % 100 == 0:
yield output(str(value + ones) + suffix)
else:
yield output(str(value) + str(ones) + suffix)
value = None
elif current in self.tens:
tens = self.tens[current]
if value is None:
value = tens
elif isinstance(value, str):
value = str(value) + str(tens)
else:
if value % 100 == 0:
value += tens
else:
value = str(value) + str(tens)
elif current in self.tens_suffixed:
# ordinal or cardinal; yield the number right away
tens, suffix = self.tens_suffixed[current]
if value is None:
yield output(str(tens) + suffix)
elif isinstance(value, str):
yield output(str(value) + str(tens) + suffix)
else:
if value % 100 == 0:
yield output(str(value + tens) + suffix)
else:
yield output(str(value) + str(tens) + suffix)
elif current in self.multipliers:
multiplier = self.multipliers[current]
if value is None:
value = multiplier
elif isinstance(value, str) or value == 0:
f = to_fraction(value)
p = f * multiplier if f is not None else None
if f is not None and p.denominator == 1:
value = p.numerator
else:
yield output(value)
value = multiplier
else:
before = value // 1000 * 1000
residual = value % 1000
value = before + residual * multiplier
elif current in self.multipliers_suffixed:
multiplier, suffix = self.multipliers_suffixed[current]
if value is None:
yield output(str(multiplier) + suffix)
elif isinstance(value, str):
f = to_fraction(value)
p = f * multiplier if f is not None else None
if f is not None and p.denominator == 1:
yield output(str(p.numerator) + suffix)
else:
yield output(value)
yield output(str(multiplier) + suffix)
else: # int
before = value // 1000 * 1000
residual = value % 1000
value = before + residual * multiplier
yield output(str(value) + suffix)
value = None
elif current in self.preceding_prefixers:
# apply prefix (positive, minus, etc.) if it precedes a number
if value is not None:
yield output(value)
if next in self.words or next_is_numeric:
prefix = self.preceding_prefixers[current]
else:
yield output(current)
elif current in self.following_prefixers:
# apply prefix (dollars, cents, etc.) only after a number
if value is not None:
prefix = self.following_prefixers[current]
yield output(value)
else:
yield output(current)
elif current in self.suffixers:
# apply suffix symbols (percent -> '%')
if value is not None:
suffix = self.suffixers[current]
if isinstance(suffix, dict):
if next in suffix:
yield output(str(value) + suffix[next])
skip = True
else:
yield output(value)
yield output(current)
else:
yield output(str(value) + suffix)
else:
yield output(current)
elif current in self.specials:
if next not in self.words and not next_is_numeric:
# apply special handling only if the next word can be numeric
if value is not None:
yield output(value)
yield output(current)
elif current == "and":
# ignore "and" after hundreds, thousands, etc.
if prev not in self.multipliers:
if value is not None:
yield output(value)
yield output(current)
elif current == "double" or current == "triple":
if next in self.ones or next in self.zeros:
repeats = 2 if current == "double" else 3
ones = self.ones.get(next, 0)
value = str(value or "") + str(ones) * repeats
skip = True
else:
if value is not None:
yield output(value)
yield output(current)
elif current == "point":
if next in self.decimals or next_is_numeric:
value = str(value or "") + "."
else:
# should all have been covered at this point
raise ValueError(f"Unexpected token: {current}")
else:
# all should have been covered at this point
raise ValueError(f"Unexpected token: {current}")
if value is not None:
yield output(value)
def preprocess(self, s: str):
# replace "<number> and a half" with "<number> point five"
results = []
segments = re.split(r"\band\s+a\s+half\b", s)
for i, segment in enumerate(segments):
if len(segment.strip()) == 0:
continue
if i == len(segments) - 1:
results.append(segment)
else:
results.append(segment)
last_word = segment.rsplit(maxsplit=2)[-1]
if last_word in self.decimals or last_word in self.multipliers:
results.append("point five")
else:
results.append("and a half")
s = " ".join(results)
# put a space at number/letter boundary
s = re.sub(r"([a-z])([0-9])", r"\1 \2", s)
s = re.sub(r"([0-9])([a-z])", r"\1 \2", s)
# but remove spaces which could be a suffix
s = re.sub(r"([0-9])\s+(st|nd|rd|th|s)\b", r"\1\2", s)
return s
def postprocess(self, s: str):
def combine_cents(m: Match):
try:
currency = m.group(1)
integer = m.group(2)
cents = int(m.group(3))
return f"{currency}{integer}.{cents:02d}"
except ValueError:
return m.string
def extract_cents(m: Match):
try:
return f"¢{int(m.group(1))}"
except ValueError:
return m.string
# apply currency postprocessing; "$2 and ¢7" -> "$2.07"
s = re.sub(r"([€£$])([0-9]+) (?:and )?¢([0-9]{1,2})\b", combine_cents, s)
s = re.sub(r"[€£$]0.([0-9]{1,2})\b", extract_cents, s)
# write "one(s)" instead of "1(s)", just for the readability
s = re.sub(r"\b1(s?)\b", r"one\1", s)
return s
def __call__(self, s: str):
s = self.preprocess(s)
s = " ".join(word for word in self.process_words(s.split()) if word is not None)
s = self.postprocess(s)
return s
class EnglishSpellingNormalizer:
"""
Applies British-American spelling mappings as listed in [1].
[1] https://www.tysto.com/uk-us-spelling-list.html
"""
def __init__(self):
mapping_path = os.path.join(os.path.dirname(__file__), "english.json")
self.mapping = json.load(open(mapping_path))
def __call__(self, s: str):
return " ".join(self.mapping.get(word, word) for word in s.split())
class EnglishTextNormalizer:
def __init__(self):
self.ignore_patterns = r"\b(hmm|mm|mhm|mmm|uh|um)\b"
self.replacers = {
# common contractions
r"\bwon't\b": "will not",
r"\bcan't\b": "can not",
r"\blet's\b": "let us",
r"\bain't\b": "aint",
r"\by'all\b": "you all",
r"\bwanna\b": "want to",
r"\bgotta\b": "got to",
r"\bgonna\b": "going to",
r"\bi'ma\b": "i am going to",
r"\bimma\b": "i am going to",
r"\bwoulda\b": "would have",
r"\bcoulda\b": "could have",
r"\bshoulda\b": "should have",
r"\bma'am\b": "madam",
# contractions in titles/prefixes
r"\bmr\b": "mister ",
r"\bmrs\b": "missus ",
r"\bst\b": "saint ",
r"\bdr\b": "doctor ",
r"\bprof\b": "professor ",
r"\bcapt\b": "captain ",
r"\bgov\b": "governor ",
r"\bald\b": "alderman ",
r"\bgen\b": "general ",
r"\bsen\b": "senator ",
r"\brep\b": "representative ",
r"\bpres\b": "president ",
r"\brev\b": "reverend ",
r"\bhon\b": "honorable ",
r"\basst\b": "assistant ",
r"\bassoc\b": "associate ",
r"\blt\b": "lieutenant ",
r"\bcol\b": "colonel ",
r"\bjr\b": "junior ",
r"\bsr\b": "senior ",
r"\besq\b": "esquire ",
# prefect tenses, ideally it should be any past participles, but it's harder..
r"'d been\b": " had been",
r"'s been\b": " has been",
r"'d gone\b": " had gone",
r"'s gone\b": " has gone",
r"'d done\b": " had done", # "'s done" is ambiguous
r"'s got\b": " has got",
# general contractions
r"n't\b": " not",
r"'re\b": " are",
r"'s\b": " is",
r"'d\b": " would",
r"'ll\b": " will",
r"'t\b": " not",
r"'ve\b": " have",
r"'m\b": " am",
}
self.standardize_numbers = EnglishNumberNormalizer()
self.standardize_spellings = EnglishSpellingNormalizer()
def __call__(self, s: str):
s = s.lower()
s = re.sub(r"[<\[][^>\]]*[>\]]", "", s) # remove words between brackets
s = re.sub(r"\(([^)]+?)\)", "", s) # remove words between parenthesis
s = re.sub(self.ignore_patterns, "", s)
s = re.sub(r"\s+'", "'", s) # standardize when there's a space before an apostrophe
for pattern, replacement in self.replacers.items():
s = re.sub(pattern, replacement, s)
s = re.sub(r"(\d),(\d)", r"\1\2", s) # remove commas between digits
s = re.sub(r"\.([^0-9]|$)", r" \1", s) # remove periods not followed by numbers
s = remove_symbols_and_diacritics(s, keep=".%$¢€£") # keep some symbols for numerics
s = self.standardize_numbers(s)
s = self.standardize_spellings(s)
# now remove prefix/suffix symbols that are not preceded/followed by numbers
s = re.sub(r"[.$¢€£]([^0-9])", r" \1", s)
s = re.sub(r"([^0-9])%", r"\1 ", s)
s = re.sub(r"\s+", " ", s) # replace any successive whitespace characters with a space
return s

331
whisperx/tokenizer.py
View File

@ -1,331 +0,0 @@
import os
from dataclasses import dataclass
from functools import lru_cache
from typing import List, Optional, Tuple, Union
import numpy as np
import torch
from transformers import GPT2TokenizerFast
LANGUAGES = {
"en": "english",
"zh": "chinese",
"de": "german",
"es": "spanish",
"ru": "russian",
"ko": "korean",
"fr": "french",
"ja": "japanese",
"pt": "portuguese",
"tr": "turkish",
"pl": "polish",
"ca": "catalan",
"nl": "dutch",
"ar": "arabic",
"sv": "swedish",
"it": "italian",
"id": "indonesian",
"hi": "hindi",
"fi": "finnish",
"vi": "vietnamese",
"he": "hebrew",
"uk": "ukrainian",
"el": "greek",
"ms": "malay",
"cs": "czech",
"ro": "romanian",
"da": "danish",
"hu": "hungarian",
"ta": "tamil",
"no": "norwegian",
"th": "thai",
"ur": "urdu",
"hr": "croatian",
"bg": "bulgarian",
"lt": "lithuanian",
"la": "latin",
"mi": "maori",
"ml": "malayalam",
"cy": "welsh",
"sk": "slovak",
"te": "telugu",
"fa": "persian",
"lv": "latvian",
"bn": "bengali",
"sr": "serbian",
"az": "azerbaijani",
"sl": "slovenian",
"kn": "kannada",
"et": "estonian",
"mk": "macedonian",
"br": "breton",
"eu": "basque",
"is": "icelandic",
"hy": "armenian",
"ne": "nepali",
"mn": "mongolian",
"bs": "bosnian",
"kk": "kazakh",
"sq": "albanian",
"sw": "swahili",
"gl": "galician",
"mr": "marathi",
"pa": "punjabi",
"si": "sinhala",
"km": "khmer",
"sn": "shona",
"yo": "yoruba",
"so": "somali",
"af": "afrikaans",
"oc": "occitan",
"ka": "georgian",
"be": "belarusian",
"tg": "tajik",
"sd": "sindhi",
"gu": "gujarati",
"am": "amharic",
"yi": "yiddish",
"lo": "lao",
"uz": "uzbek",
"fo": "faroese",
"ht": "haitian creole",
"ps": "pashto",
"tk": "turkmen",
"nn": "nynorsk",
"mt": "maltese",
"sa": "sanskrit",
"lb": "luxembourgish",
"my": "myanmar",
"bo": "tibetan",
"tl": "tagalog",
"mg": "malagasy",
"as": "assamese",
"tt": "tatar",
"haw": "hawaiian",
"ln": "lingala",
"ha": "hausa",
"ba": "bashkir",
"jw": "javanese",
"su": "sundanese",
}
# language code lookup by name, with a few language aliases
TO_LANGUAGE_CODE = {
**{language: code for code, language in LANGUAGES.items()},
"burmese": "my",
"valencian": "ca",
"flemish": "nl",
"haitian": "ht",
"letzeburgesch": "lb",
"pushto": "ps",
"panjabi": "pa",
"moldavian": "ro",
"moldovan": "ro",
"sinhalese": "si",
"castilian": "es",
}
@dataclass(frozen=True)
class Tokenizer:
"""A thin wrapper around `GPT2TokenizerFast` providing quick access to special tokens"""
tokenizer: "GPT2TokenizerFast"
language: Optional[str]
sot_sequence: Tuple[int]
def encode(self, text, **kwargs):
return self.tokenizer.encode(text, **kwargs)
def decode(self, token_ids: Union[int, List[int], np.ndarray, torch.Tensor], **kwargs):
return self.tokenizer.decode(token_ids, **kwargs)
def decode_with_timestamps(self, tokens) -> str:
"""
Timestamp tokens are above the special tokens' id range and are ignored by `decode()`.
This method decodes given tokens with timestamps tokens annotated, e.g. "<|1.08|>".
"""
outputs = [[]]
for token in tokens:
if token >= self.timestamp_begin:
timestamp = f"<|{(token - self.timestamp_begin) * 0.02:.2f}|>"
outputs.append(timestamp)
outputs.append([])
else:
outputs[-1].append(token)
outputs = [s if isinstance(s, str) else self.tokenizer.decode(s) for s in outputs]
return "".join(outputs)
@property
@lru_cache()
def eot(self) -> int:
return self.tokenizer.eos_token_id
@property
@lru_cache()
def sot(self) -> int:
return self._get_single_token_id("<|startoftranscript|>")
@property
@lru_cache()
def sot_lm(self) -> int:
return self._get_single_token_id("<|startoflm|>")
@property
@lru_cache()
def sot_prev(self) -> int:
return self._get_single_token_id("<|startofprev|>")
@property
@lru_cache()
def no_speech(self) -> int:
return self._get_single_token_id("<|nospeech|>")
@property
@lru_cache()
def no_timestamps(self) -> int:
return self._get_single_token_id("<|notimestamps|>")
@property
@lru_cache()
def timestamp_begin(self) -> int:
return self.tokenizer.all_special_ids[-1] + 1
@property
@lru_cache()
def language_token(self) -> int:
"""Returns the token id corresponding to the value of the `language` field"""
if self.language is None:
raise ValueError(f"This tokenizer does not have language token configured")
additional_tokens = dict(
zip(
self.tokenizer.additional_special_tokens,
self.tokenizer.additional_special_tokens_ids,
)
)
candidate = f"<|{self.language}|>"
if candidate in additional_tokens:
return additional_tokens[candidate]
raise KeyError(f"Language {self.language} not found in tokenizer.")
@property
@lru_cache()
def all_language_tokens(self) -> Tuple[int]:
result = []
for token, token_id in zip(
self.tokenizer.additional_special_tokens,
self.tokenizer.additional_special_tokens_ids,
):
if token.strip("<|>") in LANGUAGES:
result.append(token_id)
return tuple(result)
@property
@lru_cache()
def all_language_codes(self) -> Tuple[str]:
return tuple(self.decode([l]).strip("<|>") for l in self.all_language_tokens)
@property
@lru_cache()
def sot_sequence_including_notimestamps(self) -> Tuple[int]:
return tuple(list(self.sot_sequence) + [self.no_timestamps])
@property
@lru_cache()
def non_speech_tokens(self) -> Tuple[int]:
"""
Returns the list of tokens to suppress in order to avoid any speaker tags or non-speech
annotations, to prevent sampling texts that are not actually spoken in the audio, e.g.
- ♪♪♪
- ( SPEAKING FOREIGN LANGUAGE )
- [DAVID] Hey there,
keeping basic punctuations like commas, periods, question marks, exclamation points, etc.
"""
symbols = list("\"#()*+/:;<=>@[\\]^_`{|}~「」『』")
symbols += "<< >> <<< >>> -- --- -( -[ (' (\" (( )) ((( ))) [[ ]] {{ }} ♪♪ ♪♪♪".split()
# symbols that may be a single token or multiple tokens depending on the tokenizer.
# In case they're multiple tokens, suppress the first token, which is safe because:
# These are between U+2640 and U+267F miscellaneous symbols that are okay to suppress
# in generations, and in the 3-byte UTF-8 representation they share the first two bytes.
miscellaneous = set("♩♪♫♬♭♮♯")
assert all(0x2640 <= ord(c) <= 0x267F for c in miscellaneous)
# allow hyphens "-" and single quotes "'" between words, but not at the beginning of a word
result = {self.tokenizer.encode(" -")[0], self.tokenizer.encode(" '")[0]}
for symbol in symbols + list(miscellaneous):
for tokens in [self.tokenizer.encode(symbol), self.tokenizer.encode(" " + symbol)]:
if len(tokens) == 1 or symbol in miscellaneous:
result.add(tokens[0])
return tuple(sorted(result))
def _get_single_token_id(self, text) -> int:
tokens = self.tokenizer.encode(text)
assert len(tokens) == 1, f"{text} is not encoded as a single token"
return tokens[0]
@lru_cache(maxsize=None)
def build_tokenizer(name: str = "gpt2"):
os.environ["TOKENIZERS_PARALLELISM"] = "false"
path = os.path.join(os.path.dirname(__file__), "assets", name)
tokenizer = GPT2TokenizerFast.from_pretrained(path)
specials = [
"<|startoftranscript|>",
*[f"<|{lang}|>" for lang in LANGUAGES.keys()],
"<|translate|>",
"<|transcribe|>",
"<|startoflm|>",
"<|startofprev|>",
"<|nospeech|>",
"<|notimestamps|>",
]
tokenizer.add_special_tokens(dict(additional_special_tokens=specials))
return tokenizer
@lru_cache(maxsize=None)
def get_tokenizer(
multilingual: bool,
*,
task: Optional[str] = None, # Literal["transcribe", "translate", None]
language: Optional[str] = None,
) -> Tokenizer:
if language is not None:
language = language.lower()
if language not in LANGUAGES:
if language in TO_LANGUAGE_CODE:
language = TO_LANGUAGE_CODE[language]
else:
raise ValueError(f"Unsupported language: {language}")
if multilingual:
tokenizer_name = "multilingual"
task = task or "transcribe"
language = language or "en"
else:
tokenizer_name = "gpt2"
task = None
language = None
tokenizer = build_tokenizer(name=tokenizer_name)
all_special_ids: List[int] = tokenizer.all_special_ids
sot: int = all_special_ids[1]
translate: int = all_special_ids[-6]
transcribe: int = all_special_ids[-5]
langs = tuple(LANGUAGES.keys())
sot_sequence = [sot]
if language is not None:
sot_sequence.append(sot + 1 + langs.index(language))
if task is not None:
sot_sequence.append(transcribe if task == "transcribe" else translate)
return Tokenizer(tokenizer=tokenizer, language=language, sot_sequence=tuple(sot_sequence))

776
whisperx/transcribe.py
View File

@ -1,602 +1,58 @@
import argparse
import os
import warnings
from typing import List, Optional, Tuple, Union, Iterator, TYPE_CHECKING
from typing import TYPE_CHECKING, Optional, Tuple, Union
import numpy as np
import torch
import tqdm
from .audio import SAMPLE_RATE, N_FRAMES, HOP_LENGTH, CHUNK_LENGTH, pad_or_trim, log_mel_spectrogram, load_audio
from .alignment import load_align_model, align, get_trellis, backtrack, merge_repeats, merge_words
from .decoding import DecodingOptions, DecodingResult
from .diarize import assign_word_speakers, Segment
from .tokenizer import LANGUAGES, TO_LANGUAGE_CODE, get_tokenizer
from .utils import exact_div, format_timestamp, optional_int, optional_float, str2bool, interpolate_nans, write_txt, write_vtt, write_srt, write_ass, write_tsv
from .vad import Binarize
import pandas as pd
if TYPE_CHECKING:
from .model import Whisper
from whisper.tokenizer import LANGUAGES, TO_LANGUAGE_CODE
from whisper.utils import (
optional_float,
optional_int,
str2bool,
)
from .utils import get_writer
def transcribe(
model: "Whisper",
audio: Union[str, np.ndarray, torch.Tensor],
*,
verbose: Optional[bool] = None,
temperature: Union[float, Tuple[float, ...]] = (0.0, 0.2, 0.4, 0.6, 0.8, 1.0),
compression_ratio_threshold: Optional[float] = 2.4,
logprob_threshold: Optional[float] = -1.0,
no_speech_threshold: Optional[float] = 0.6,
condition_on_previous_text: bool = False, # turn off by default due to errors it causes
mel: np.ndarray = None,
**decode_options,
):
"""
Transcribe an audio file using Whisper
Parameters
----------
model: Whisper
The Whisper model instance
audio: Union[str, np.ndarray, torch.Tensor]
The path to the audio file to open, or the audio waveform
verbose: bool
Whether to display the text being decoded to the console. If True, displays all the details,
If False, displays minimal details. If None, does not display anything
temperature: Union[float, Tuple[float, ...]]
Temperature for sampling. It can be a tuple of temperatures, which will be successfully used
upon failures according to either `compression_ratio_threshold` or `logprob_threshold`.
compression_ratio_threshold: float
If the gzip compression ratio is above this value, treat as failed
logprob_threshold: float
If the average log probability over sampled tokens is below this value, treat as failed
no_speech_threshold: float
If the no_speech probability is higher than this value AND the average log probability
over sampled tokens is below `logprob_threshold`, consider the segment as silent
condition_on_previous_text: bool
if True, the previous output of the model is provided as a prompt for the next window;
disabling may make the text inconsistent across windows, but the model becomes less prone to
getting stuck in a failure loop, such as repetition looping or timestamps going out of sync.
decode_options: dict
Keyword arguments to construct `DecodingOptions` instances
Returns
-------
A dictionary containing the resulting text ("text") and segment-level details ("segments"), and
the spoken language ("language"), which is detected when `decode_options["language"]` is None.
"""
dtype = torch.float16 if decode_options.get("fp16", True) else torch.float32
if model.device == torch.device("cpu"):
if torch.cuda.is_available():
warnings.warn("Performing inference on CPU when CUDA is available")
if dtype == torch.float16:
warnings.warn("FP16 is not supported on CPU; using FP32 instead")
dtype = torch.float32
if dtype == torch.float32:
decode_options["fp16"] = False
if mel is None:
mel = log_mel_spectrogram(audio)
if decode_options.get("language", None) is None:
if not model.is_multilingual:
decode_options["language"] = "en"
else:
if verbose:
print("Detecting language using up to the first 30 seconds. Use `--language` to specify the language")
segment = pad_or_trim(mel, N_FRAMES).to(model.device).to(dtype)
_, probs = model.detect_language(segment)
decode_options["language"] = max(probs, key=probs.get)
if verbose is not None:
print(f"Detected language: {LANGUAGES[decode_options['language']].title()}")
language = decode_options["language"]
task = decode_options.get("task", "transcribe")
tokenizer = get_tokenizer(model.is_multilingual, language=language, task=task)
def decode_with_fallback(segment: torch.Tensor) -> DecodingResult:
temperatures = [temperature] if isinstance(temperature, (int, float)) else temperature
decode_result = None
for t in temperatures:
kwargs = {**decode_options}
if t > 0:
# disable beam_size and patience when t > 0
kwargs.pop("beam_size", None)
kwargs.pop("patience", None)
else:
# disable best_of when t == 0
kwargs.pop("best_of", None)
options = DecodingOptions(**kwargs, temperature=t)
decode_result = model.decode(segment, options)
needs_fallback = False
if compression_ratio_threshold is not None and decode_result.compression_ratio > compression_ratio_threshold:
needs_fallback = True # too repetitive
if logprob_threshold is not None and decode_result.avg_logprob < logprob_threshold:
needs_fallback = True # average log probability is too low
if not needs_fallback:
break
return decode_result
seek = 0
input_stride = exact_div(
N_FRAMES, model.dims.n_audio_ctx
) # mel frames per output token: 2
time_precision = (
input_stride * HOP_LENGTH / SAMPLE_RATE
) # time per output token: 0.02 (seconds)
all_tokens = []
all_segments = []
prompt_reset_since = 0
initial_prompt = decode_options.pop("initial_prompt", None) or []
if initial_prompt:
initial_prompt = tokenizer.encode(" " + initial_prompt.strip())
all_tokens.extend(initial_prompt)
def add_segment(
*, start: float, end: float, text_tokens: torch.Tensor, result: DecodingResult
):
text = tokenizer.decode([token for token in text_tokens if token < tokenizer.eot])
if len(text.strip()) == 0: # skip empty text output
return
all_segments.append(
{
"id": len(all_segments),
"seek": seek,
"start": start,
"end": end,
"text": text,
"tokens": text_tokens.tolist(),
"temperature": result.temperature,
"avg_logprob": result.avg_logprob,
"compression_ratio": result.compression_ratio,
"no_speech_prob": result.no_speech_prob,
}
)
if verbose:
print(f"[{format_timestamp(start)} --> {format_timestamp(end)}] {text}")
# show the progress bar when verbose is False (otherwise the transcribed text will be printed)
num_frames = mel.shape[-1]
previous_seek_value = seek
with tqdm.tqdm(total=num_frames, unit='frames', disable=verbose is not False) as pbar:
while seek < num_frames:
timestamp_offset = float(seek * HOP_LENGTH / SAMPLE_RATE)
segment = pad_or_trim(mel[:, seek:], N_FRAMES).to(model.device).to(dtype)
segment_duration = segment.shape[-1] * HOP_LENGTH / SAMPLE_RATE
decode_options["prompt"] = all_tokens[prompt_reset_since:]
result: DecodingResult = decode_with_fallback(segment)
tokens = torch.tensor(result.tokens)
if no_speech_threshold is not None:
# no voice activity check
should_skip = result.no_speech_prob > no_speech_threshold
if logprob_threshold is not None and result.avg_logprob > logprob_threshold:
# don't skip if the logprob is high enough, despite the no_speech_prob
should_skip = False
if should_skip:
seek += segment.shape[-1] # fast-forward to the next segment boundary
continue
timestamp_tokens: torch.Tensor = tokens.ge(tokenizer.timestamp_begin)
consecutive = torch.where(timestamp_tokens[:-1] & timestamp_tokens[1:])[0].add_(1)
if len(consecutive) > 0: # if the output contains two consecutive timestamp tokens
last_slice = 0
for current_slice in consecutive:
sliced_tokens = tokens[last_slice:current_slice]
start_timestamp_position = (
sliced_tokens[0].item() - tokenizer.timestamp_begin
)
end_timestamp_position = (
sliced_tokens[-1].item() - tokenizer.timestamp_begin
)
# clamp end-time to at least be 1 frame after start-time
end_timestamp_position = max(end_timestamp_position, start_timestamp_position + time_precision)
add_segment(
start=timestamp_offset + start_timestamp_position * time_precision,
end=timestamp_offset + end_timestamp_position * time_precision,
text_tokens=sliced_tokens[1:-1],
result=result,
)
last_slice = current_slice
last_timestamp_position = (
tokens[last_slice - 1].item() - tokenizer.timestamp_begin
)
seek += last_timestamp_position * input_stride
all_tokens.extend(tokens[: last_slice + 1].tolist())
else:
duration = segment_duration
timestamps = tokens[timestamp_tokens.nonzero().flatten()]
if len(timestamps) > 0 and timestamps[-1].item() != tokenizer.timestamp_begin:
# no consecutive timestamps but it has a timestamp; use the last one.
# single timestamp at the end means no speech after the last timestamp.
last_timestamp_position = timestamps[-1].item() - tokenizer.timestamp_begin
duration = last_timestamp_position * time_precision
add_segment(
start=timestamp_offset,
end=timestamp_offset + duration,
text_tokens=tokens,
result=result,
)
seek += segment.shape[-1]
all_tokens.extend(tokens.tolist())
if not condition_on_previous_text or result.temperature > 0.5:
# do not feed the prompt tokens if a high temperature was used
prompt_reset_since = len(all_tokens)
# update progress bar
pbar.update(min(num_frames, seek) - previous_seek_value)
previous_seek_value = seek
return dict(text=tokenizer.decode(all_tokens[len(initial_prompt):]), segments=all_segments, language=language)
def merge_chunks(segments, chunk_size=CHUNK_LENGTH):
"""
Merge VAD segments into larger segments of approximately size ~CHUNK_LENGTH.
TODO: Make sure VAD segment isn't too long, otherwise it will cause OOM when input to alignment model
TODO: Or sliding window alignment model over long segment.
"""
curr_end = 0
merged_segments = []
seg_idxs = []
speaker_idxs = []
assert chunk_size > 0
binarize = Binarize(max_duration=chunk_size)
segments = binarize(segments)
segments_list = []
for speech_turn in segments.get_timeline():
segments_list.append(Segment(speech_turn.start, speech_turn.end, "UNKNOWN"))
assert segments_list, "segments_list is empty."
# Make sur the starting point is the start of the segment.
curr_start = segments_list[0].start
for seg in segments_list:
if seg.end - curr_start > chunk_size and curr_end-curr_start > 0:
merged_segments.append({
"start": curr_start,
"end": curr_end,
"segments": seg_idxs,
})
curr_start = seg.start
seg_idxs = []
speaker_idxs = []
curr_end = seg.end
seg_idxs.append((seg.start, seg.end))
speaker_idxs.append(seg.speaker)
# add final
merged_segments.append({
"start": curr_start,
"end": curr_end,
"segments": seg_idxs,
})
return merged_segments
def transcribe_with_vad(
model: "Whisper",
audio: Union[str, np.ndarray, torch.Tensor],
vad_pipeline,
mel = None,
verbose: Optional[bool] = None,
**kwargs
):
"""
Transcribe per VAD segment
"""
if mel is None:
mel = log_mel_spectrogram(audio)
prev = 0
output = {"segments": []}
vad_segments = vad_pipeline(audio)
# merge segments to approx 30s inputs to make whisper most appropraite
vad_segments = merge_chunks(vad_segments)
for sdx, seg_t in enumerate(vad_segments):
if verbose:
print(f"~~ Transcribing VAD chunk: ({format_timestamp(seg_t['start'])} --> {format_timestamp(seg_t['end'])}) ~~")
seg_f_start, seg_f_end = int(seg_t["start"] * SAMPLE_RATE / HOP_LENGTH), int(seg_t["end"] * SAMPLE_RATE / HOP_LENGTH)
local_f_start, local_f_end = seg_f_start - prev, seg_f_end - prev
mel = mel[:, local_f_start:] # seek forward
prev = seg_f_start
local_mel = mel[:, :local_f_end-local_f_start]
result = transcribe(model, audio, mel=local_mel, verbose=verbose, **kwargs)
seg_t["text"] = result["text"]
output["segments"].append(
{
"start": seg_t["start"],
"end": seg_t["end"],
"language": result["language"],
"text": result["text"],
"seg-text": [x["text"] for x in result["segments"]],
"seg-start": [x["start"] for x in result["segments"]],
"seg-end": [x["end"] for x in result["segments"]],
}
)
output["language"] = output["segments"][0]["language"]
return output
def transcribe_with_vad_parallel(
model: "Whisper",
audio: Union[str, np.ndarray, torch.Tensor],
vad_pipeline,
mel = None,
verbose: Optional[bool] = None,
batch_size = -1,
**kwargs
):
"""
Transcribe per VAD segment
"""
if mel is None:
mel = log_mel_spectrogram(audio)
vad_segments = vad_pipeline(audio)
# merge segments to approx 30s inputs to make whisper most appropraite
vad_segments = merge_chunks(vad_segments)
################################
### START of parallelization ###
################################
# pad mel to a same length
start_seconds = [i['start'] for i in vad_segments]
end_seconds = [i['end'] for i in vad_segments]
duration_list = np.array(end_seconds) - np.array(start_seconds)
max_length = round(30 / (HOP_LENGTH / SAMPLE_RATE))
offset_list = np.array(start_seconds)
chunks = []
for start_ts, end_ts in zip(start_seconds, end_seconds):
start_ts = round(start_ts / (HOP_LENGTH / SAMPLE_RATE))
end_ts = round(end_ts / (HOP_LENGTH / SAMPLE_RATE))
chunk = mel[:, start_ts:end_ts]
chunk = torch.nn.functional.pad(chunk, (0, max_length-chunk.shape[-1]))
chunks.append(chunk)
mel_chunk = torch.stack(chunks, dim=0).to(model.device)
# using 'decode_options1': only support single temperature decoding (no fallbacks)
# result_list2 = model.decode(mel_chunk, decode_options1)
# prepare DecodingOptions
temperatures = kwargs.pop("temperature", None)
compression_ratio_threshold = kwargs.pop("compression_ratio_threshold", None)
logprob_threshold = kwargs.pop("logprob_threshold", None)
no_speech_threshold = kwargs.pop("no_speech_threshold", None)
condition_on_previous_text = kwargs.pop("condition_on_previous_text", None)
initial_prompt = kwargs.pop("initial_prompt", None)
t = 0 # TODO: does not upport temperature sweeping
if t > 0:
# disable beam_size and patience when t > 0
kwargs.pop("beam_size", None)
kwargs.pop("patience", None)
else:
# disable best_of when t == 0
kwargs.pop("best_of", None)
options = DecodingOptions(**kwargs, temperature=t)
mel_chunk_batches = torch.split(mel_chunk, split_size_or_sections=batch_size)
decode_result = []
for mel_chunk_batch in mel_chunk_batches:
decode_result.extend(model.decode(mel_chunk_batch, options))
##############################
### END of parallelization ###
##############################
# post processing: get segments rfom batch-decoded results
input_stride = exact_div(
N_FRAMES, model.dims.n_audio_ctx
) # mel frames per output token: 2
language = kwargs["language"]
task = kwargs["task"]
tokenizer = get_tokenizer(model.is_multilingual, language=language, task=task)
output = post_process_results(
vad_segments,
decode_result,
duration_list,
offset_list,
input_stride,
language,
tokenizer,
no_speech_threshold=no_speech_threshold,
logprob_threshold=logprob_threshold,
verbose=verbose)
return output
def post_process_results(
vad_segments,
result_list,
duration_list,
offset_list,
input_stride,
language,
tokenizer,
no_speech_threshold = None,
logprob_threshold = None,
verbose: Optional[bool] = None,
):
seek = 0
time_precision = (
input_stride * HOP_LENGTH / SAMPLE_RATE
) # time per output token: 0.02 (seconds)
all_tokens = []
all_segments = []
output = {"segments": []}
def add_segment(
*, start: float, end: float, text_tokens: torch.Tensor, result: DecodingResult
):
text = tokenizer.decode([token for token in text_tokens if token < tokenizer.eot])
if len(text.strip()) == 0: # skip empty text output
return
all_segments.append(
{
"id": len(all_segments),
"seek": seek,
"start": start,
"end": end,
"text": text,
"tokens": text_tokens.tolist(),
"temperature": result.temperature,
"avg_logprob": result.avg_logprob,
"compression_ratio": result.compression_ratio,
"no_speech_prob": result.no_speech_prob,
}
)
if verbose:
print(f"[{format_timestamp(start)} --> {format_timestamp(end)}] {text}")
# process the output
for seg_t, result, segment_duration, timestamp_offset in zip(vad_segments, result_list, duration_list, offset_list):
all_tokens = []
all_segments = []
# segment_duration = segment.shape[-1] * HOP_LENGTH / SAMPLE_RATE
segment_shape = int(segment_duration / (HOP_LENGTH / SAMPLE_RATE))
tokens = torch.tensor(result.tokens)
if no_speech_threshold is not None:
# no voice activity check
should_skip = result.no_speech_prob > no_speech_threshold
if logprob_threshold is not None and result.avg_logprob > logprob_threshold:
# don't skip if the logprob is high enough, despite the no_speech_prob
should_skip = False
if should_skip:
seek += segment_shape # fast-forward to the next segment boundary
continue
timestamp_tokens: torch.Tensor = tokens.ge(tokenizer.timestamp_begin)
consecutive = torch.where(timestamp_tokens[:-1] & timestamp_tokens[1:])[0].add_(1)
if len(consecutive) > 0: # if the output contains two consecutive timestamp tokens
last_slice = 0
for current_slice in consecutive:
sliced_tokens = tokens[last_slice:current_slice]
start_timestamp_position = (
sliced_tokens[0].item() - tokenizer.timestamp_begin
)
end_timestamp_position = (
sliced_tokens[-1].item() - tokenizer.timestamp_begin
)
add_segment(
start=timestamp_offset + start_timestamp_position * time_precision,
end=timestamp_offset + end_timestamp_position * time_precision,
text_tokens=sliced_tokens[1:-1],
result=result,
)
last_slice = current_slice
last_timestamp_position = (
tokens[last_slice - 1].item() - tokenizer.timestamp_begin
)
seek += last_timestamp_position * input_stride
all_tokens.extend(tokens[: last_slice + 1].tolist())
else:
duration = segment_duration
timestamps = tokens[timestamp_tokens.nonzero().flatten()]
if len(timestamps) > 0 and timestamps[-1].item() != tokenizer.timestamp_begin:
# no consecutive timestamps but it has a timestamp; use the last one.
# single timestamp at the end means no speech after the last timestamp.
last_timestamp_position = timestamps[-1].item() - tokenizer.timestamp_begin
duration = last_timestamp_position * time_precision
add_segment(
start=timestamp_offset,
end=timestamp_offset + duration,
text_tokens=tokens,
result=result,
)
seek += segment_shape
all_tokens.extend(tokens.tolist())
result = dict(text=tokenizer.decode(all_tokens), segments=all_segments, language=language)
output["segments"].append(
{
"start": seg_t["start"],
"end": seg_t["end"],
"language": result["language"],
"text": result["text"],
"seg-text": [x["text"] for x in result["segments"]],
"seg-start": [x["start"] for x in result["segments"]],
"seg-end": [x["end"] for x in result["segments"]],
}
)
output["language"] = output["segments"][0]["language"]
return output
from .asr import transcribe, transcribe_with_vad
from .alignment import load_align_model, align
from .diarize import DiarizationPipeline
from .vad import load_vad_model
def cli():
from . import available_models
from whisper import available_models
# fmt: off
parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument("audio", nargs="+", type=str, help="audio file(s) to transcribe")
parser.add_argument("--model", default="small", choices=available_models(), help="name of the Whisper model to use")
parser.add_argument("--model_dir", type=str, default=None, help="the path to save model files; uses ~/.cache/whisper by default")
parser.add_argument("--device", default="cuda" if torch.cuda.is_available() else "cpu", help="device to use for PyTorch inference")
# alignment params
parser.add_argument("--align_model", default=None, help="Name of phoneme-level ASR model to do alignment")
parser.add_argument("--align_extend", default=2, type=float, help="Seconds before and after to extend the whisper segments for alignment")
parser.add_argument("--align_from_prev", default=True, type=bool, help="Whether to clip the alignment start time of current segment to the end time of the last aligned word of the previous segment")
parser.add_argument("--interpolate_method", default="nearest", choices=["nearest", "linear", "ignore"], help="For word .srt, method to assign timestamps to non-aligned words, or merge them into neighbouring.")
# vad params
parser.add_argument("--vad_filter", action="store_true", help="Whether to first perform VAD filtering to target only transcribe within VAD. Produces more accurate alignment + timestamp, requires more GPU memory & compute.")
parser.add_argument("--parallel_bs", default=-1, type=int, help="Enable parallel transcribing if > 1")
# diarization params
parser.add_argument("--diarize", action="store_true", help="Apply diarization to assign speaker labels to each segment/word")
parser.add_argument("--min_speakers", default=None, type=int)
parser.add_argument("--max_speakers", default=None, type=int)
# output save params
parser.add_argument("--output_dir", "-o", type=str, default=".", help="directory to save the outputs")
parser.add_argument("--output_type", default="all", choices=["all", "srt", "srt-word", "vtt", "txt", "tsv", "ass", "ass-char", "pickle", "vad"], help="File type for desired output save")
parser.add_argument("--output_format", "-f", type=str, default="all", choices=["all", "srt", "srt-word", "vtt", "txt", "tsv", "ass", "ass-char", "pickle", "vad"], help="format of the output file; if not specified, all available formats will be produced")
parser.add_argument("--verbose", type=str2bool, default=True, help="whether to print out the progress and debug messages")
parser.add_argument("--task", type=str, default="transcribe", choices=["transcribe", "translate"], help="whether to perform X->X speech recognition ('transcribe') or X->English translation ('translate')")
parser.add_argument("--language", type=str, default=None, choices=sorted(LANGUAGES.keys()) + sorted([k.title() for k in TO_LANGUAGE_CODE.keys()]), help="language spoken in the audio, specify None to perform language detection")
# alignment params
parser.add_argument("--align_model", default=None, help="Name of phoneme-level ASR model to do alignment")
parser.add_argument("--align_extend", default=2, type=float, help="Seconds before and after to extend the whisper segments for alignment (if not using VAD).")
parser.add_argument("--align_from_prev", default=True, type=bool, help="Whether to clip the alignment start time of current segment to the end time of the last aligned word of the previous segment (if not using VAD)")
parser.add_argument("--interpolate_method", default="nearest", choices=["nearest", "linear", "ignore"], help="For word .srt, method to assign timestamps to non-aligned words, or merge them into neighbouring.")
parser.add_argument("--no_align", action='store_true', help="Do not perform phoneme alignment")
# vad params
parser.add_argument("--vad_filter", action="store_true", help="Whether to pre-segment audio with VAD, highly recommended! Produces more accurate alignment + timestamp see WhisperX paper https://arxiv.org/abs/2303.00747")
parser.add_argument("--vad_onset", type=float, default=0.767, help="Onset threshold for VAD (see pyannote.audio)")
parser.add_argument("--vad_offset", type=float, default=0.363, help="Offset threshold for VAD (see pyannote.audio).")
# diarization params
parser.add_argument("--diarize", action="store_true", help="Apply diarization to assign speaker labels to each segment/word")
parser.add_argument("--min_speakers", default=None, type=int)
parser.add_argument("--max_speakers", default=None, type=int)
parser.add_argument("--temperature", type=float, default=0, help="temperature to use for sampling")
parser.add_argument("--best_of", type=optional_int, default=5, help="number of candidates when sampling with non-zero temperature")
parser.add_argument("--beam_size", type=optional_int, default=5, help="number of beams in beam search, only applicable when temperature is zero")
@ -605,160 +61,114 @@ def cli():
parser.add_argument("--suppress_tokens", type=str, default="-1", help="comma-separated list of token ids to suppress during sampling; '-1' will suppress most special characters except common punctuations")
parser.add_argument("--initial_prompt", type=str, default=None, help="optional text to provide as a prompt for the first window.")
parser.add_argument("--condition_on_previous_text", type=str2bool, default=False, help="if True, provide the previous output of the model as a prompt for the next window; disabling may make the text inconsistent across windows, but the model becomes less prone to getting stuck in a failure loop")
parser.add_argument("--condition_on_previous_text", type=str2bool, default=True, help="if True, provide the previous output of the model as a prompt for the next window; disabling may make the text inconsistent across windows, but the model becomes less prone to getting stuck in a failure loop")
parser.add_argument("--fp16", type=str2bool, default=True, help="whether to perform inference in fp16; True by default")
parser.add_argument("--temperature_increment_on_fallback", type=optional_float, default=0.2, help="temperature to increase when falling back when the decoding fails to meet either of the thresholds below")
parser.add_argument("--compression_ratio_threshold", type=optional_float, default=2.4, help="if the gzip compression ratio is higher than this value, treat the decoding as failed")
parser.add_argument("--logprob_threshold", type=optional_float, default=-1.0, help="if the average log probability is lower than this value, treat the decoding as failed")
parser.add_argument("--no_speech_threshold", type=optional_float, default=0.6, help="if the probability of the <|nospeech|> token is higher than this value AND the decoding has failed due to `logprob_threshold`, consider the segment as silence")
parser.add_argument("--word_timestamps", type=str2bool, default=False, help="(experimental) extract word-level timestamps and refine the results based on them")
parser.add_argument("--prepend_punctuations", type=str, default="\"\'“¿([{-", help="if word_timestamps is True, merge these punctuation symbols with the next word")
parser.add_argument("--append_punctuations", type=str, default="\"\'.。,，!！?？:：”)]}、", help="if word_timestamps is True, merge these punctuation symbols with the previous word")
parser.add_argument("--threads", type=optional_int, default=0, help="number of threads used by torch for CPU inference; supercedes MKL_NUM_THREADS/OMP_NUM_THREADS")
parser.add_argument("--hf_token", type=str, default=None, help="Hugging Face Access Token to access PyAnnote gated models")
parser.add_argument("--model_flush", action="store_true", help="Flush memory of each stage after use, more GPU memory efficient, but slower when there are multiple audio files")
# fmt: on
args = parser.parse_args().__dict__
model_name: str = args.pop("model")
model_dir: str = args.pop("model_dir")
output_dir: str = args.pop("output_dir")
output_type: str = args.pop("output_type")
output_format: str = args.pop("output_format")
device: str = args.pop("device")
model_flush: bool = args.pop("model_flush")
os.makedirs(output_dir, exist_ok=True)
align_model: str = args.pop("align_model")
align_extend: float = args.pop("align_extend")
align_from_prev: bool = args.pop("align_from_prev")
interpolate_method: bool = args.pop("interpolate_method")
interpolate_method: str = args.pop("interpolate_method")
no_align: bool = args.pop("no_align")
hf_token: str = args.pop("hf_token")
vad_filter: bool = args.pop("vad_filter")
parallel_bs: int = args.pop("parallel_bs")
vad_onset: float = args.pop("vad_onset")
vad_offset: float = args.pop("vad_offset")
diarize: bool = args.pop("diarize")
min_speakers: int = args.pop("min_speakers")
max_speakers: int = args.pop("max_speakers")
vad_pipeline = None
if vad_filter:
if hf_token is None:
print("Warning, no huggingface token used, needs to be saved in environment variable, otherwise will throw error loading VAD model...")
from pyannote.audio import Inference, Model
vad_pipeline = Inference(
Model.from_pretrained("pyannote/segmentation",
use_auth_token=hf_token),
pre_aggregation_hook=lambda segmentation: segmentation,
use_auth_token=hf_token,
device=torch.device(device),
)
diarize_pipeline = None
if diarize:
if hf_token is None:
print("Warning, no --hf_token used, needs to be saved in environment variable, otherwise will throw error loading diarization model...")
from pyannote.audio import Pipeline
diarize_pipeline = Pipeline.from_pretrained("pyannote/speaker-diarization@2.1",
use_auth_token=hf_token)
from pyannote.audio import Model, Pipeline
vad_model = load_vad_model(torch.device(device), vad_onset, vad_offset, use_auth_token=hf_token)
else:
vad_model = None
if diarize:
if hf_token is None:
print("Warning, no --hf_token used, needs to be saved in environment variable, otherwise will throw error loading diarization model...")
diarize_model = DiarizationPipeline(use_auth_token=hf_token)
else:
diarize_model = None
if no_align:
align_model, align_metadata = None, None
else:
align_language = args["language"] if args["language"] is not None else "en" # default to loading english if not specified
align_model, align_metadata = load_align_model(align_language, device, model_name=align_model)
os.makedirs(output_dir, exist_ok=True)
if model_name.endswith(".en") and args["language"] not in {"en", "English"}:
if args["language"] is not None:
warnings.warn(f'{model_name} is an English-only model but receipted "{args["language"]}"; using English instead.')
warnings.warn(
f"{model_name} is an English-only model but receipted '{args['language']}'; using English instead."
)
args["language"] = "en"
temperature = args.pop("temperature")
temperature_increment_on_fallback = args.pop("temperature_increment_on_fallback")
if temperature_increment_on_fallback is not None:
temperature = tuple(np.arange(temperature, 1.0 + 1e-6, temperature_increment_on_fallback))
if (increment := args.pop("temperature_increment_on_fallback")) is not None:
temperature = tuple(np.arange(temperature, 1.0 + 1e-6, increment))
else:
temperature = [temperature]
threads = args.pop("threads")
if threads > 0:
if (threads := args.pop("threads")) > 0:
torch.set_num_threads(threads)
from . import load_model
from whisper import load_model
model = load_model(model_name, device=device, download_root=model_dir)
align_language = args["language"] if args["language"] is not None else "en" # default to loading english if not specified
align_model, align_metadata = load_align_model(align_language, device, model_name=align_model)
writer = get_writer(output_format, output_dir)
for audio_path in args.pop("audio"):
if vad_filter:
if parallel_bs > 1:
print("Performing VAD and parallel transcribing ...")
result = transcribe_with_vad_parallel(model, audio_path, vad_pipeline, temperature=temperature, batch_size=parallel_bs, **args)
else:
print("Performing VAD...")
result = transcribe_with_vad(model, audio_path, vad_pipeline, temperature=temperature, **args)
if vad_model is not None:
print("Performing VAD...")
result = transcribe_with_vad(model, audio_path, vad_model, temperature=temperature, **args)
else:
print("Performing transcription...")
result = transcribe(model, audio_path, temperature=temperature, **args)
if result["language"] != align_metadata["language"]:
# load new language
print(f"New language found ({result['language']})! Previous was ({align_metadata['language']}), loading new alignment model for new language...")
align_model, align_metadata = load_align_model(result["language"], device)
if align_model is not None:
if result["language"] != align_metadata["language"]:
# load new language
print(f"New language found ({result['language']})! Previous was ({align_metadata['language']}), loading new alignment model for new language...")
align_model, align_metadata = load_align_model(result["language"], device)
result = align(result["segments"], align_model, align_metadata, audio_path, device,
extend_duration=align_extend, start_from_previous=align_from_prev, interpolate_method=interpolate_method)
# if diarize_model is not None:
# diarize_segments = diarize_model(audio_path, min_speakers=min_speakers, max_speakers=max_speakers)
# results_segments, word_segments = assign_word_speakers(diarize_segments, )
writer(result, audio_path)
print("Performing alignment...")
result_aligned = align(result["segments"], align_model, align_metadata, audio_path, device,
extend_duration=align_extend, start_from_previous=align_from_prev, interpolate_method=interpolate_method)
audio_basename = os.path.basename(audio_path)
if diarize:
print("Performing diarization...")
diarize_segments = diarize_pipeline(audio_path, min_speakers=min_speakers, max_speakers=max_speakers)
diarize_df = pd.DataFrame(diarize_segments.itertracks(yield_label=True))
diarize_df['start'] = diarize_df[0].apply(lambda x: x.start)
diarize_df['end'] = diarize_df[0].apply(lambda x: x.end)
# assumes each utterance is single speaker (needs fix)
result_segments, word_segments = assign_word_speakers(diarize_df, result_aligned["segments"], fill_nearest=True)
result_aligned["segments"] = result_segments
result_aligned["word_segments"] = word_segments
# save TXT
if output_type in ["txt", "all"]:
with open(os.path.join(output_dir, audio_basename + ".txt"), "w", encoding="utf-8") as txt:
write_txt(result_aligned["segments"], file=txt)
# save VTT
if output_type in ["vtt", "all"]:
with open(os.path.join(output_dir, audio_basename + ".vtt"), "w", encoding="utf-8") as vtt:
write_vtt(result_aligned["segments"], file=vtt)
# save SRT
if output_type in ["srt", "all"]:
with open(os.path.join(output_dir, audio_basename + ".srt"), "w", encoding="utf-8") as srt:
write_srt(result_aligned["segments"], file=srt)
# save TSV
if output_type in ["tsv", "all"]:
with open(os.path.join(output_dir, audio_basename + ".tsv"), "w", encoding="utf-8") as srt:
write_tsv(result_aligned["segments"], file=srt)
# save SRT word-level
if output_type in ["srt-word", "all"]:
# save per-word SRT
with open(os.path.join(output_dir, audio_basename + ".word.srt"), "w", encoding="utf-8") as srt:
write_srt(result_aligned["word_segments"], file=srt)
# save ASS
if output_type in ["ass", "all"]:
with open(os.path.join(output_dir, audio_basename + ".ass"), "w", encoding="utf-8") as ass:
write_ass(result_aligned["segments"], file=ass)
# # save ASS character-level
if output_type in ["ass-char"]:
with open(os.path.join(output_dir, audio_basename + ".char.ass"), "w", encoding="utf-8") as ass:
write_ass(result_aligned["segments"], file=ass, resolution="char")
# save word tsv
if output_type in ["pickle"]:
exp_fp = os.path.join(output_dir, audio_basename + ".pkl")
pd.DataFrame(result_aligned["segments"]).to_pickle(exp_fp)
# save word tsv
if output_type in ["vad"]:
exp_fp = os.path.join(output_dir, audio_basename + ".sad")
wrd_segs = pd.concat([x["word-segments"] for x in result_aligned["segments"]])[['start','end']]
wrd_segs.to_csv(exp_fp, sep='\t', header=None, index=False)
if __name__ == "__main__":
cli()
cli()

140
whisperx/utils.py
View File

@ -4,48 +4,12 @@ from typing import Callable, TextIO, Iterator, Tuple
import pandas as pd
import numpy as np
def exact_div(x, y):
assert x % y == 0
return x // y
def str2bool(string):
str2val = {"True": True, "False": False}
if string in str2val:
return str2val[string]
def interpolate_nans(x, method='nearest'):
if x.notnull().sum() > 1:
return x.interpolate(method=method).ffill().bfill()
else:
raise ValueError(f"Expected one of {set(str2val.keys())}, got {string}")
def optional_int(string):
return None if string == "None" else int(string)
def optional_float(string):
return None if string == "None" else float(string)
def compression_ratio(text) -> float:
text_bytes = text.encode("utf-8")
return len(text_bytes) / len(zlib.compress(text_bytes))
def format_timestamp(seconds: float, always_include_hours: bool = False, decimal_marker: str = '.'):
assert seconds >= 0, "non-negative timestamp expected"
milliseconds = round(seconds * 1000.0)
hours = milliseconds // 3_600_000
milliseconds -= hours * 3_600_000
minutes = milliseconds // 60_000
milliseconds -= minutes * 60_000
seconds = milliseconds // 1_000
milliseconds -= seconds * 1_000
hours_marker = f"{hours:02d}:" if always_include_hours or hours > 0 else ""
return f"{hours_marker}{minutes:02d}:{seconds:02d}{decimal_marker}{milliseconds:03d}"
return x.ffill().bfill()
def write_txt(transcript: Iterator[dict], file: TextIO):
for segment in transcript:
@ -250,8 +214,92 @@ def write_ass(transcript: Iterator[dict],
file.write(ass_str)
def interpolate_nans(x, method='nearest'):
if x.notnull().sum() > 1:
return x.interpolate(method=method).ffill().bfill()
else:
return x.ffill().bfill()
from whisper.utils import SubtitlesWriter, ResultWriter, WriteTXT, WriteVTT, WriteSRT, WriteTSV, WriteJSON, format_timestamp
class WriteASS(ResultWriter):
extension: str = "ass"
def write_result(self, result: dict, file: TextIO):
write_ass(result["segments"], file, resoltuion="word")
class WriteASSchar(ResultWriter):
extension: str = "ass"
def write_result(self, result: dict, file: TextIO):
write_ass(result["segments"], file, resoltuion="char")
class WritePickle(ResultWriter):
extension: str = "ass"
def write_result(self, result: dict, file: TextIO):
pd.DataFrame(result["segments"]).to_pickle(file)
class WriteSRTWord(ResultWriter):
extension: str = ".word.srt"
always_include_hours: bool = True
decimal_marker: str = ","
def iterate_result(self, result: dict):
for segment in result["word_segments"]:
segment_start = self.format_timestamp(segment["start"])
segment_end = self.format_timestamp(segment["end"])
segment_text = segment["text"].strip().replace("-->", "->")
if word_timings := segment.get("words", None):
all_words = [timing["word"] for timing in word_timings]
all_words[0] = all_words[0].strip() # remove the leading space, if any
last = segment_start
for i, this_word in enumerate(word_timings):
start = self.format_timestamp(this_word["start"])
end = self.format_timestamp(this_word["end"])
if last != start:
yield last, start, segment_text
yield start, end, "".join(
[
f"<u>{word}</u>" if j == i else word
for j, word in enumerate(all_words)
]
)
last = end
if last != segment_end:
yield last, segment_end, segment_text
else:
yield segment_start, segment_end, segment_text
def write_result(self, result: dict, file: TextIO):
for i, (start, end, text) in enumerate(self.iterate_result(result), start=1):
print(f"{i}\n{start} --> {end}\n{text}\n", file=file, flush=True)
def format_timestamp(self, seconds: float):
return format_timestamp(
seconds=seconds,
always_include_hours=self.always_include_hours,
decimal_marker=self.decimal_marker,
)
def get_writer(output_format: str, output_dir: str) -> Callable[[dict, TextIO], None]:
writers = {
"txt": WriteTXT,
"vtt": WriteVTT,
"srt": WriteSRT,
"tsv": WriteTSV,
# "json": WriteJSON,
"ass": WriteASS,
# "ass-char": WriteASSchar,
# "pickle": WritePickle,
"srt-word": WriteSRTWord,
}
if output_format == "all":
all_writers = [writer(output_dir) for writer in writers.values()]
def write_all(result: dict, file: TextIO):
for writer in all_writers:
writer(result, file)
return write_all
return writers[output_format](output_dir)

128
whisperx/vad.py
View File

@ -1,10 +1,32 @@
import os
import pandas as pd
import numpy as np
from pyannote.core import Annotation, Segment, SlidingWindowFeature, Timeline
import torch
from typing import Optional, Callable, Union, Text
from pyannote.audio.core.io import AudioFile
from pyannote.core import Annotation, Segment, SlidingWindowFeature
from pyannote.audio.pipelines.utils import PipelineModel
from pyannote.audio import Model, Pipeline
from pyannote.audio.pipelines import VoiceActivityDetection
from .diarize import Segment as SegmentX
from typing import List, Tuple, Optional
def load_vad_model(device, vad_onset, vad_offset, use_auth_token=None):
vad_model = Model.from_pretrained("pyannote/segmentation", use_auth_token=use_auth_token)
hyperparameters = {"onset": vad_onset,
"offset": vad_offset,
"min_duration_on": 0.1,
"min_duration_off": 0.1}
vad_pipeline = VoiceActivitySegmentation(segmentation=vad_model, device=torch.device(device))
vad_pipeline.instantiate(hyperparameters)
return vad_pipeline
class Binarize:
"""Binarize detection scores using hysteresis thresholding
"""Binarize detection scores using hysteresis thresholding, with min-cut operation
to ensure not segments are longer than max_duration.
Parameters
----------
onset : float, optional
@ -28,6 +50,9 @@ class Binarize:
Gregory Gelly and Jean-Luc Gauvain. "Minimum Word Error Training of
RNN-based Voice Activity Detection", InterSpeech 2015.
Modified by Max Bain to include WhisperX's min-cut operation
https://arxiv.org/abs/2303.00747
Pyannote-audio
"""
@ -136,6 +161,51 @@ class Binarize:
return active
class VoiceActivitySegmentation(VoiceActivityDetection):
def __init__(
self,
segmentation: PipelineModel = "pyannote/segmentation",
fscore: bool = False,
use_auth_token: Union[Text, None] = None,
**inference_kwargs,
):
super().__init__(segmentation=segmentation, fscore=fscore, use_auth_token=use_auth_token, **inference_kwargs)
def apply(self, file: AudioFile, hook: Optional[Callable] = None) -> Annotation:
"""Apply voice activity detection
Parameters
----------
file : AudioFile
Processed file.
hook : callable, optional
Hook called after each major step of the pipeline with the following
signature: hook("step_name", step_artefact, file=file)
Returns
-------
speech : Annotation
Speech regions.
"""
# setup hook (e.g. for debugging purposes)
hook = self.setup_hook(file, hook=hook)
# apply segmentation model (only if needed)
# output shape is (num_chunks, num_frames, 1)
if self.training:
if self.CACHED_SEGMENTATION in file:
segmentations = file[self.CACHED_SEGMENTATION]
else:
segmentations = self._segmentation(file)
file[self.CACHED_SEGMENTATION] = segmentations
else:
segmentations: SlidingWindowFeature = self._segmentation(file)
return segmentations
def merge_vad(vad_arr, pad_onset=0.0, pad_offset=0.0, min_duration_off=0.0, min_duration_on=0.0):
active = Annotation()
@ -157,21 +227,49 @@ def merge_vad(vad_arr, pad_onset=0.0, pad_offset=0.0, min_duration_off=0.0, min_
active_segs = pd.DataFrame([x['segment'] for x in active['content']])
return active_segs
def merge_chunks(segments, chunk_size):
"""
Merge operation described in paper
"""
curr_end = 0
merged_segments = []
seg_idxs = []
speaker_idxs = []
assert chunk_size > 0
binarize = Binarize(max_duration=chunk_size)
segments = binarize(segments)
segments_list = []
for speech_turn in segments.get_timeline():
segments_list.append(SegmentX(speech_turn.start, speech_turn.end, "UNKNOWN"))
assert segments_list, "segments_list is empty."
# Make sur the starting point is the start of the segment.
curr_start = segments_list[0].start
for seg in segments_list:
if seg.end - curr_start > chunk_size and curr_end-curr_start > 0:
merged_segments.append({
"start": curr_start,
"end": curr_end,
"segments": seg_idxs,
})
curr_start = seg.start
seg_idxs = []
speaker_idxs = []
curr_end = seg.end
seg_idxs.append((seg.start, seg.end))
speaker_idxs.append(seg.speaker)
# add final
merged_segments.append({
"start": curr_start,
"end": curr_end,
"segments": seg_idxs,
})
return merged_segments
if __name__ == "__main__":
# from pyannote.audio import Inference
# hook = lambda segmentation: segmentation
# inference = Inference("pyannote/segmentation", pre_aggregation_hook=hook)
# audio = "/tmp/11962.wav"
# scores = inference(audio)
# binarize = Binarize(max_duration=15)
# anno = binarize(scores)
# res = []
# for ann in anno.get_timeline():
# res.append((ann.start, ann.end))
# res = pd.DataFrame(res)
# res[2] = res[1] - res[0]
import pandas as pd
input_fp = "tt298650_sync.wav"
df = pd.read_csv(f"/work/maxbain/tmp/{input_fp}.sad", sep=" ", header=None)