gpt-omni commited on
Commit
b7df447
1 Parent(s): 0d7b9d9
Files changed (1) hide show
  1. app.py +8 -4
app.py CHANGED
@@ -312,6 +312,12 @@ def process_audio(audio):
312
  if filepath is None:
313
  return OUT_RATE, np.zeros((100, OUT_CHANNELS), dtype=np.int16)
314
 
 
 
 
 
 
 
315
  cnt = 0
316
  tik = time.time()
317
  for chunk in run_AT_batch_stream(filepath):
@@ -327,15 +333,13 @@ def process_audio(audio):
327
  return OUT_RATE, audio_data.astype(np.int16)
328
 
329
 
330
- demo = gr.Interface(
 
331
  process_audio,
332
  inputs=gr.Audio(type="filepath", label="Microphone"),
333
  outputs=[gr.Audio(label="Response", streaming=streaming_output, autoplay=True)],
334
  title="Chat Mini-Omni Demo",
335
  live=True,
336
  )
337
-
338
-
339
- if __name__ == '__main__':
340
  demo.queue()
341
  demo.launch()
 
312
  if filepath is None:
313
  return OUT_RATE, np.zeros((100, OUT_CHANNELS), dtype=np.int16)
314
 
315
+ if not streaming_output:
316
+ chunk = run_AT_batch_stream(filepath)
317
+ audio_data = np.frombuffer(chunk, dtype=np.int16)
318
+ audio_data = audio_data.reshape(-1, OUT_CHANNELS)
319
+ return OUT_RATE, audio_data.astype(np.int16)
320
+
321
  cnt = 0
322
  tik = time.time()
323
  for chunk in run_AT_batch_stream(filepath):
 
333
  return OUT_RATE, audio_data.astype(np.int16)
334
 
335
 
336
+ if __name__ == '__main__':
337
+ demo = gr.Interface(
338
  process_audio,
339
  inputs=gr.Audio(type="filepath", label="Microphone"),
340
  outputs=[gr.Audio(label="Response", streaming=streaming_output, autoplay=True)],
341
  title="Chat Mini-Omni Demo",
342
  live=True,
343
  )
 
 
 
344
  demo.queue()
345
  demo.launch()