Python interface to the RNNoise VAD(Voice Activity Detection) component inside webrtc
| CI | status |
|---|---|
| pip builds |  |
cibuildwheel |
 |
- A compiler with C++11 support
- Pip 10+ or CMake >= 3.4 (or 3.8+ on Windows, which was the first version to support VS 2015)
- Ninja or Pip 10+
- Numpy
pip install py-webrtcrnnvad
RNNoise - BSD PyBind - BSD
import py_webrtcrnnvad
audio, fs = read_wave("audio.wav")
frames = frame_generator(10, audio, 48000)
frame = np.frombuffer(frames[0].bytes, dtype=np.short)
vad = py_webrtcrnnvad.RNNVAD()
vad.processFrame(frame.astype(np.float32))