speech-detector

NodeJS library providing VAD (voice activity detection) or more specifically speech activity detection. This library processes a raw stream of PCM audio data and emits a stream of PCM audio data segements that contain speech. This library leverages the Silero model for speech detection along with the ONNX framework.

Install

npm install --save "speech-detector"

Usage

import { SpeechDetector } from "speech-detector";

// Create a SpeechDetector using all default values.
const speechDetector = await SpeechDetector.create();

const speechSegments = await speechDetector.process(audioStream);

for await (const segement of speechSegments) {
  console.log(`Received speech segement: ${segement}`);
}