ValueError: operands could not be broadcast together with shapes

[Hackebein]

After running for some hours it suddenly crashed(?) and was stuck there. Clicking Stop and Start in the UI fixed it temporare.

Conversion took: 0.07958483695983887
Conversion took: 0.07854104042053223
Conversion took: 0.08368110656738281
Conversion took: 0.08663725852966309
Conversion took: 0.08125996589660645
Conversion took: 0.0883171558380127
Conversion took: 0.18883609771728516
Conversion took: 2.0105159282684326
Model generation logic took longer than specified latency.
Conversion took: 0.05684185028076172
conversion_process_target: stopped
Exception in thread Thread-4:
Traceback (most recent call last):
  File "C:\Users\User\AppData\Local\Programs\Python\Python310\lib\threading.py", line 1016, in _bootstrap_inner
    self.run()
  File "C:\Users\User\Projects\GradioMVL\conversion_thread.py", line 62, in run
    out = self.voice_conversion.run(wav, HDW_FRAMES_PER_BUFFER)
  File "C:\Users\User\Projects\GradioMVL\voice_conversion.py", line 132, in run
    return self.run_linear_crossfade(wav, HDW_FRAMES_PER_BUFFER)
  File "C:\Users\User\Projects\GradioMVL\voice_conversion.py", line 154, in run_linear_crossfade
    out[-(HDW_FRAMES_PER_BUFFER + self._fade_samples) : -HDW_FRAMES_PER_BUFFER] * self._linear_fade_in
ValueError: operands could not be broadcast together with shapes (0,) (441,)

[ferret99gt]

What were the settings in use? This was caused by the spike in GPU processing just above, where it jumped from the normal 86ms to 2 full seconds. The audio buffers overflowed as a result.

[Hackebein]

Input latenxy 300 Crossface Methid linear Voice File yara