This repo contains build instructions and example of usage for voice activity detection. For now only C++ is supported, though in future python bindings should be added.
This repo is inspired by https://github.com/wiseman/py-webrtcvad Which, unfortunately, uses outdated version of WebRTC code base.
In order to build you will need g++, cmake. Only linux supported for now.
git clone --recursive https://github.com/StcInc/webrtc_vad
cd webrtc_vad
./build.sh
The output should be in build folder - test app build/vad_test
cd build/
./vad_test
cd ..
cat out.txt
Test app processes test_wav.wav and outputs vad predictions to out.txt
Each line signifies prediction for 5 ms interval. These are raw predictions without any smoothing.
# https://abseil.io
git clone https://github.com/abseil/abseil-cpp.git
# webrtc https://webrtc.org
git clone https://webrtc.googlesource.com/src