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

missing word timestamps

Browse Source
This commit is contained in:
Max Bain
2023-01-24 16:37:19 +00:00
parent d1600e5b0f
commit eec6d1f8d8
2 changed files with 26 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
whisperx/transcribe.py
View File

@ -330,8 +330,8 @@ def align(
aligned_segments = []
prev_t2 = 0
sdx = 0
for segment in transcript:
aligned_subsegments = []
while True:
segment_align_success = False
@ -505,7 +505,7 @@ def align(
else:
word_level = None
aligned_segments.append(
aligned_subsegments.append(
{
"text": segment["seg-text"][sub_seg_idx],
"start": seg_start_actual,
@ -515,10 +515,7 @@ def align(
}
)
if "language" in segment:
aligned_segments[-1]["language"] = segment["language"]
print(f"[{format_timestamp(aligned_segments[-1]['start'])} --> {format_timestamp(aligned_segments[-1]['end'])}] {aligned_segments[-1]['text']}")
aligned_subsegments[-1]["language"] = segment["language"]
char_level = {
"start": [],
@ -555,12 +552,14 @@ def align(
# reset prev_t2 due to drifting issues
if not segment_align_success:
prev_t2 = 0
start = interpolate_nans(pd.DataFrame(aligned_subsegments)["start"], method=interpolate_method)
end = interpolate_nans(pd.DataFrame(aligned_subsegments)["end"], method=interpolate_method)
for idx, seg in enumerate(aligned_subsegments):
seg['start'] = start.iloc[idx]
seg['end'] = end.iloc[idx]
# shift segment index by amount of sub-segments
if "seg-text" in segment:
sdx += len(segment["seg-text"])
else:
sdx += 1
aligned_segments += aligned_subsegments
# create word level segments for .srt
word_seg = []