SnJForever commited on
Commit
883bac9
β€’
1 Parent(s): 0a75ccd
Files changed (1) hide show
  1. app.py +15 -5
app.py CHANGED
@@ -59,8 +59,18 @@ def on_prompt_template_change(prompt_template):
59
  if not isinstance(prompt_template, str): return
60
  return prompt_templates[prompt_template]
61
 
 
 
 
 
 
 
 
 
 
 
62
  # UNCOMMENT TO USE WHISPER
63
- def transcribe(aud_inp, whisper_lang):
64
  if aud_inp is None:
65
  return ""
66
  text = WHISPER_MODEL.transcribe(aud_inp)
@@ -327,12 +337,12 @@ with gr.Blocks(css=css) as demo:
327
  with gr.Column(scale=3):
328
  btn_submit = gr.Button("Submit")
329
  total_tokens_str = gr.Markdown(elem_id="total_tokens_str")
 
 
 
 
330
  with gr.Row():
331
  btn_clear_conversation = gr.Button("πŸ”ƒ Start New Conversation")
332
- # with gr.Row():
333
- # audio_comp = gr.Microphone(source="microphone", type="filepath", label="Just say it!",
334
- # interactive=True, streaming=False)
335
- # audio_comp.change(transcribe, inputs=[audio_comp, WHISPER_DETECT_LANG], outputs=[input_message])
336
 
337
 
338
  # gr.HTML('''<br><br><br><center>You can duplicate this Space to skip the queue:<a href="https://huggingface.co/spaces/anzorq/chatgpt-demo?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a><br>
 
59
  if not isinstance(prompt_template, str): return
60
  return prompt_templates[prompt_template]
61
 
62
+ def lan_detector(audio_file):
63
+ print('reading the audio file')
64
+ audio = whisper.load_audio(audio_file)
65
+ audio = whisper.pad_or_trim(audio)
66
+ mel = whisper.log_mel_spectrogram(audio).to(WHISPER_MODEL.device)
67
+ _, probs = WHISPER_MODEL.detect_language(mel)
68
+ if max(probs, key=probs.get) == 'en':
69
+ return True
70
+ return False
71
+
72
  # UNCOMMENT TO USE WHISPER
73
+ def transcribe(aud_inp):
74
  if aud_inp is None:
75
  return ""
76
  text = WHISPER_MODEL.transcribe(aud_inp)
 
337
  with gr.Column(scale=3):
338
  btn_submit = gr.Button("Submit")
339
  total_tokens_str = gr.Markdown(elem_id="total_tokens_str")
340
+ with gr.Row():
341
+ audio_comp = gr.Microphone(source="microphone", type="filepath", label="Just say it!",
342
+ interactive=True, streaming=False)
343
+ audio_comp.change(transcribe, inputs=[audio_comp], outputs=[input_message])
344
  with gr.Row():
345
  btn_clear_conversation = gr.Button("πŸ”ƒ Start New Conversation")
 
 
 
 
346
 
347
 
348
  # gr.HTML('''<br><br><br><center>You can duplicate this Space to skip the queue:<a href="https://huggingface.co/spaces/anzorq/chatgpt-demo?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a><br>