kotoba-tech
/

kotoba-whisper-v1.0-faster

Automatic Speech Recognition

Model card Files Files and versions Community

asahi417 commited on May 8, 2024

Commit

7ec3fd2

·

verified ·

1 Parent(s): 1482df5

Update benchmark.sh

Files changed (1) hide show

benchmark.sh +4 -5

benchmark.sh CHANGED Viewed

@@ -8,16 +8,15 @@ ffmpeg -i kotoba-whisper-eval/audio/manzai3.mp3 -ar 16000 -ac 1 -c:a pcm_s16le k
 # cache the model
 python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster")'
 SECONDS=0
-python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); segments=model.transcribe("kotoba-whisper-eval/audio/long_interview_1.wav", language="ja", chunk_length=15, condition_on_previous_text=False); for segment in segments:print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))'
 TIME_INTERVIEW=$SECONDS
 SECONDS=0
-python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); model.transcribe("kotoba-whisper-eval/audio/manzai1.wav", language="ja", chunk_length=15, condition_on_previous_text=False)"'
 TIME_MANZAI1=$SECONDS
 SECONDS=0
-python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); model.transcribe("kotoba-whisper-eval/audio/manzai2.wav", language="ja", chunk_length=15, condition_on_previous_text=False)"'
 TIME_MANZAI2=$SECONDS
 SECONDS=0
-python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); model.transcribe("kotoba-whisper-eval/audio/manzai3.wav", language="ja", chunk_length=15, condition_on_previous_text=False)"'
 TIME_MANZAI3=$SECONDS

 # cache the model
 python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster")'
 SECONDS=0
+python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); print(["[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text) for segment in model.transcribe("kotoba-whisper-eval/audio/long_interview_1.wav", language="ja", chunk_length=15, condition_on_previous_text=False)[0]])'
 TIME_INTERVIEW=$SECONDS
 SECONDS=0
+python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); print(["[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text) for segment in model.transcribe("kotoba-whisper-eval/audio/manzai1.wav", language="ja", chunk_length=15, condition_on_previous_text=False)[0]])'
 TIME_MANZAI1=$SECONDS
 SECONDS=0
+python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); print(["[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text) for segment in model.transcribe("kotoba-whisper-eval/audio/manzai2.wav", language="ja", chunk_length=15, condition_on_previous_text=False)[0]])'
 TIME_MANZAI2=$SECONDS
 SECONDS=0
+python -c 'from faster_whisper import WhisperModel; model = WhisperModel("kotoba-tech/kotoba-whisper-v1.0-faster"); print(["[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text) for segment in model.transcribe("kotoba-whisper-eval/audio/manzai3.wav", language="ja", chunk_length=15, condition_on_previous_text=False)[0]])'
 TIME_MANZAI3=$SECONDS