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

Merge pull request #473 from sorgfresser/fix-faster-whisper-threads

Browse Source
This commit is contained in:
Max Bain
2023-09-19 16:53:34 -07:00
committed by GitHub
parent b1a98b78c9 0ae0d49d1d
commit ffd6167b26
2 changed files with 8 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
whisperx/asr.py
View File

@ -31,7 +31,8 @@ def load_model(whisper_arch,
vad_options=None,
model=None,
task="transcribe",
download_root=None):
download_root=None,
threads=4):
'''Load a Whisper model for inference.
Args:
whisper_arch: str - The name of the Whisper model to load.
@ -40,6 +41,7 @@ def load_model(whisper_arch,
options: dict - A dictionary of options to use for the model.
language: str - The language of the model. (use English for now)
download_root: Optional[str] - The root directory to download the model to.
threads: int - The number of cpu threads to use per worker, e.g. will be multiplied by num workers.
Returns:
A Whisper pipeline.
'''
@ -51,7 +53,8 @@ def load_model(whisper_arch,
device=device,
device_index=device_index,
compute_type=compute_type,
download_root=download_root)
download_root=download_root,
cpu_threads=threads)
if language is not None:
tokenizer = faster_whisper.tokenizer.Tokenizer(model.hf_tokenizer, model.model.is_multilingual, task=task, language=language)
else:

4
whisperx/transcribe.py
View File

@ -126,8 +126,10 @@ def cli():
else:
temperature = [temperature]
faster_whisper_threads = 4
if (threads := args.pop("threads")) > 0:
torch.set_num_threads(threads)
faster_whisper_threads = threads
asr_options = {
"beam_size": args.pop("beam_size"),
@ -157,7 +159,7 @@ def cli():
results = []
tmp_results = []
# model = load_model(model_name, device=device, download_root=model_dir)
model = load_model(model_name, device=device, device_index=device_index, compute_type=compute_type, language=args['language'], asr_options=asr_options, vad_options={"vad_onset": vad_onset, "vad_offset": vad_offset}, task=task)
model = load_model(model_name, device=device, device_index=device_index, compute_type=compute_type, language=args['language'], asr_options=asr_options, vad_options={"vad_onset": vad_onset, "vad_offset": vad_offset}, task=task, threads=faster_whisper_threads)
for audio_path in args.pop("audio"):
audio = load_audio(audio_path)