snakers4/silero-vad

Silero VAD: pre-trained enterprise-grade Voice Activity Detector

What it solves

Silero VAD is a pre-trained, enterprise-grade Voice Activity Detector (VAD) designed to accurately identify whether speech is present in an audio stream, filtering out background noise and non-speech sounds.

How it works

It uses a lightweight, pre-trained model (approximately 2MB in size) that can be run via PyTorch or ONNX Runtime. The model was trained on a massive dataset covering over 6,000 languages, allowing it to perform well across different domains and noise levels. It supports sampling rates of 8,000 Hz and 16,000 Hz.

Who it’s for

Developers building voice interfaces, IoT/edge/mobile applications, telephony automation, call-center bots, and those performing data cleaning for audio datasets.

Highlights

  • High Performance: Processes audio chunks (30+ ms) in less than 1ms on a single CPU thread.
  • Broad Language Support: Trained on 6,000+ languages for high robustness.
  • Highly Portable: Compatible with PyTorch and ONNX, enabling deployment across various platforms including C++, Rust, Go, Java, and C#.
  • Lightweight: Small model size (~2MB) makes it ideal for edge devices.
  • Permissive License: MIT license with no telemetry or vendor lock-in.

Related

  • Project
    filippogiruzzi/voice_activity_detection

    A deep learning-based Voice Activity Detection system using a 1D-ResNet and MFCC features to classify audio signals as speech or noise.

  • Project
    KoljaB/RealtimeSTT

    A Python speech-to-text library that provides fast, real-time transcription with integrated voice activity detection and wake word support.

  • Project
    QwenAudio/SenseVoice

    A high-accuracy, low-latency speech foundation model that provides multilingual speech recognition, emotion detection, and audio event identification.

  • Project
    amsehili/auditok

    A lightweight Python library for audio activity detection that splits audio streams into events using energy thresholding or WebRTC VAD.

  • Project
    FireRedTeam/FireRedASR2S

    An industrial-grade all-in-one ASR system that combines speech recognition, voice activity detection, language identification, and punctuation prediction, with specialized support for Chinese dialects.