Silero-VAD-ExecuTorch-XNNPACK

Pre-exported ExecuTorch .pte file for Silero VAD with XNNPACK backend (CPU). A lightweight (~1.2 MB) voice activity detection model that runs on any platform.

Installation

git clone https://github.com/pytorch/executorch/ ~/executorch
cd ~/executorch && ./install_executorch.sh
make silero-vad-cpu

Download

pip install huggingface_hub
huggingface-cli download younghan-meta/Silero-VAD-ExecuTorch-XNNPACK --local-dir ~/silero_vad

Run

cmake-out/examples/models/silero_vad/silero_vad_runner \
    --model_path ~/silero_vad/silero_vad.pte \
    --audio_path ~/silero_vad/poem.wav

Output shows detected speech segments with start/end times in seconds.

Export Command

python examples/models/silero_vad/export_silero_vad.py \
    --jit-model silero_vad.jit --backend xnnpack --output-dir ./silero_vad_exports

The silero_vad.jit file can be obtained from the silero-vad repository.

More Info

• Official ExecuTorch Silero VAD guide
• Silero VAD