Enhancement request - Limit audio file chunk sizes about py-webrtcvad

[jehoshua7]

@manbharae - I remembered I had a permission error when installing, but this fixed it

pip3 install --user webrtcvad

and then to run it, set the aggressive mode and specify filename

python3 example.py 0 output.wav

for agressive mode of 0, and

python3 example.py 1 output.wav

for agressive mode of 1, etc.

[jehoshua7]

how does it know where to find test.wav to process?

It's all relative. If your file test.wav is in the current path, it is

python3 example.py 1 test.wav

If your file test.wav is in some other path, it is

python3 example.py 1 ~/someotherpath/output.wav

[manbharae]

@jehoshua7 Could you tell how to process a .wav/.mp3 using py-webrtcvad?
i.e After doing pip install py-webrtcvad what do I do?

[manbharae]

I too had the installation issue. your suggestion of --user tag helped! Thank you very much for pointing it 👍 .

Question:
say my file is name test.wav
in the python3 example.py 1 output.wav how does it know where to find test.wav to process?

[wiseman]

example.py is just meant to show you how to write code that uses the VAD, not be a full-featured application. But you could try changing line 150 to use a smaller time window. Instead of

segments = vad_collector(sample_rate, 30, 300, vad, frames)

try this, which changes the time window from 300 ms to 100 ms:

segments = vad_collector(sample_rate, 30, 100, vad, frames)

[jehoshua7]

@wiseman Yes, thanks John, that was the line I was looking at last night. So, good timing, thanks. :)

[sagar2226]

@jehoshua7 how can i get audio chunks size no more than 9 seconds?

[sagar2226]

@wiseman Yes, thanks John, that was the line I was looking at last night. So, good timing, thanks. :)

@jehoshua7 how can i get audio chunks size no more than 9 seconds?
i got it in this way.. my chunks wont be no more than 9 seconds now
if len(voiced_frames) > 300:
sys.stdout.write('-(%s)' % (frame.timestamp + frame.duration))
end.append(frame.timestamp + frame.duration)
triggered = False
yield b''.join([f.bytes for f in voiced_frames])
ring_buffer.clear()
voiced_frames = []

[sagar2226]

@wiseman Yes, thanks John, that was the line I was looking at last night. So, good timing, thanks. :)

@jehoshua7 how can i get audio chunks size no more than 9 seconds?
i got it in this way.. my chunks wont be no more than 9 seconds now
Update in the else part of vad collector
if len(voiced_frames) > 300:
sys.stdout.write('-(%s)' % (frame.timestamp + frame.duration))
end.append(frame.timestamp + frame.duration)
triggered = False
yield b''.join([f.bytes for f in voiced_frames])
ring_buffer.clear()
voiced_frames = []
this code will be followed by elif num_voiced>0.9
*****************
*****************