Decibri ships two VAD implementations built in. No cloud API, no extra install, no separate model download.
'energy' mode.'silero' mode.The energy detector is a lightweight RMS computation; the Silero neural model runs locally via ONNX Runtime. The Silero ONNX model (~2.3 MB) ships inside the decibri package on both PyPI and npm. No separate download, no sherpa-onnx dependency.
import decibri
# RMS energy detector. Fast, clean-audio use cases.
mic_rms = decibri.Microphone(vad="energy")
# Silero neural model. Better accuracy under noise / music / multi-speaker.
mic_silero = decibri.Microphone(vad="silero")
# Python exposes VAD state as properties.
with mic_silero:
for chunk in mic_silero:
if mic_silero.is_speaking:
print("[speech]")const { Microphone } = require('decibri');
// RMS energy detector. Fast, clean-audio use cases.
const micRms = new Microphone({ vad: 'energy' });
// Silero neural model. Better accuracy under noise / music / multi-speaker.
const micSilero = new Microphone({ vad: 'silero' });
// Node emits 'speech' / 'silence' events.
micSilero.on('speech', () => console.log('[speech start]'));
micSilero.on('silence', () => console.log('[speech end]'));VAD configuration options are documented in full in the Python and Node.js API references.