Skip to content

thurti/vad-audio-worklet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

NPM Version

VAD AudioWorklet

AudioWorklet implementation of the vad algorithm from Moattar, Mohammad & Homayoonpoor, Mahdi .

Example Mic/File
Full Example with Debug Data

Reference

Moattar, Mohammad & Homayoonpoor, Mahdi. (2010). A simple but efficient real-time voice activity detection algorithm. European Signal Processing Conference.

https://www.researchgate.net/publication/255667085_A_simple_but_efficient_real-time_voice_activity_detection_algorithm

Install

Copy the files from /src to your public folder.

src/vad-audio-worklet.js
src/fft.js

When using npm install you need to copy the files from node_modules into your project directory. ES6 module import wouldn't work.

npm install vad-audio-worklet

Usage

Copy the files in /src to your project. Create a new AudioContext, load the vad module and connect the vad worklet node to some audio source.

More on how to use the Web Audio API see https://developer.mozilla.org/en-US/docs/Web/API/Web_Audio_API

// create AudioContext
const audioContext = new AudioContext();

// load AudioWorklet module
await audioContext.audioWorklet.addModule("/src/vad-audio-worklet.js");

// create new vad audio node
const vad = new AudioWorkletNode(audioContext, "vad", {
  outputChannelCount: [1],
  processorOptions: {
    sampleRate: audioContext.sampleRate, // sample rate of the audio input
    fftSize: 128, // optional change fft size, default: 128
  },
});

// connect worklet to some audio source
yourAudioSource.connect(vad);

// listen for messages
vad.port.onmessage = (event) => {
  const cmd = event.data["cmd"];

  if (cmd === "speech") {
    // speech detected
  }

  if (cmd === "silence") {
    // silence , cpt. obvious
  }
};

Example File

example/alert.ogg Radio Universidad Nacional de La Plata, CC BY-SA 3.0, via Wikimedia Commons

Credits

fft.js
https://github.com/indutny/fft.js/

Moattar, Mohammad & Homayoonpoor, Mahdi. (2010). A simple but efficient real-time voice activity detection algorithm. European Signal Processing Conference.
https://www.researchgate.net/publication/255667085_A_simple_but_efficient_real-time_voice_activity_detection_algorithm

MDN Docs: Web Audio API
https://developer.mozilla.org/en-US/docs/Web/API/Web_Audio_API
https://openwebdocs.org/