asr-inference

Running on Zero

ssolito commited on 10 days ago

Commit

297d08e

verified ·

1 Parent(s): f6b1fad

Update whisper.py (#9)

- Update whisper.py (e14e7976daf042a7b9fc332904bef2dda4423cc0)

Co-authored-by: Sarah Solito <ssolito@users.noreply.huggingface.co>

Files changed (1) hide show

whisper.py CHANGED Viewed

@@ -141,7 +141,7 @@ def generate(audio_path, use_v2):
             temp_mono_path = "temp_mono.wav"
             audio.export(temp_mono_path, format="wav")
             audio_path = temp_mono_path
-            task = "transcribe"
         output = transcribe_pipeline(format_audio(audio_path), task)
     clean_output = post_process_transcription(output, max_repeats=1) #check

             temp_mono_path = "temp_mono.wav"
             audio.export(temp_mono_path, format="wav")
             audio_path = temp_mono_path
+        task = "transcribe"
         output = transcribe_pipeline(format_audio(audio_path), task)
     clean_output = post_process_transcription(output, max_repeats=1) #check