microphone.js

microphone.js is a simple library that allows you to use the new Web Speech API (currently only available in Chrome) to recognize and act on voice commands in JavaScript.

How to use

Clone this repository, and spin up a simple web server to serve index.html and the JavaScript.

$ git clone https://github.com/stderr/microphone.js.git
$ cd microphone.js/
$ python -m SimpleHTTPServer

Usage

To instantiate the main Microphone object, use the create function:

require(['microphone'], function(microphone){
  var Microphone = microphone.create();
});

To activate voice processing, use the on function:

Microphone.on();

The primary usage of microphone.js is through the addCommand function. This takes two main arguments: patterns, which is assumed to be an array of regular expressions, and the callback, which is the code that will be run if any of the patterns specified match the voice input:

Microphone.addCommand({
  patterns: [new RegExp('dog', 'i')],
  callback: function() {
    alert("no, cats!");
  }
});

The callback also receives two arguments: the current Microphone instance, allowing the user to inspect the currentTranscript and other relevant information; and the matched CommandString, which exposes the matched string, and the args, which is an array of the words following the matched string. This allows you to do things like:

Microphone.addCommand({
  patterns: [new RegExp("display", "i")],
  callback: function(mic, command) {
    $("#" + command.args[0]).show();
  }
});

Thus, if you said "display foo", the DOM element with an id of foo would be shown (presuming you are using jQuery, of course).