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

torch2.0, remove compile for now, round to times to 3 decimal

Browse Source
This commit is contained in:
Max Bain
2023-05-04 20:38:13 +01:00
parent d2116b98ca
commit 4e2ac4e4e9
6 changed files with 40 additions and 34 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
whisperx/transcribe.py
View File

@ -72,7 +72,6 @@ def cli():
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 from each model after use, reduces GPU requirement but slower processing >1 audio file.")
parser.add_argument("--tmp_dir", default=None, help="Temporary directory to write audio file if input if not .wav format (only for VAD).")
# fmt: on
args = parser.parse_args().__dict__
@ -86,10 +85,6 @@ def cli():
# model_flush: bool = args.pop("model_flush")
os.makedirs(output_dir, exist_ok=True)
tmp_dir: str = args.pop("tmp_dir")
if tmp_dir is not None:
os.makedirs(tmp_dir, exist_ok=True)
align_model: str = args.pop("align_model")
interpolate_method: str = args.pop("interpolate_method")
no_align: bool = args.pop("no_align")
@ -195,7 +190,7 @@ def cli():
tmp_results = results
print(">>Performing diarization...")
results = []
diarize_model = DiarizationPipeline(use_auth_token=hf_token, device=device)
diarize_model = DiarizationPipeline(use_auth_token=hf_token)
for result, input_audio_path in tmp_results:
diarize_segments = diarize_model(input_audio_path, min_speakers=min_speakers, max_speakers=max_speakers)
results_segments, word_segments = assign_word_speakers(diarize_segments, result["segments"])