Skip to content
Implement voice commands using the Speech Recognition API
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


Implement voice commands through the Speech Recognition API


Include the script (either minified or development version):

<script src="path/to/ultvoice.min.js"></script>

This will expose a global ultvoice object which provides the API documented below.

Now, push object(s) to ultvoice's actions array:

	trigger: 'open',
	callback: function(address) {
		location = 'http://' + address;

All you need to do now is simply call ultvoice.start() to start listening for voice input! Then just say "open" and you will be redirected to Google. =]

Keep reading for the API details.


The action objects' trigger property accepts either a string or array. This way, you can specify an array of triggers for multiple languages:

	trigger: ['open', 'abra'],
	callback: function(address) {
		location = 'http://' + address;

And now saying either "open google dot com" or "abra google ponto com" (Portuguese) will redirect the page to

This feature can also be used implement synonyms (mutliple words in a given language to trigger the same callback).

The action object

Each element of the ultvoice.actions array must be an UltVoice action object. Action objects are nothing but plain JS objects which contain trigger and callback properties.

  • trigger: a word or array of words that will trigger the callback function once recognized.
  • callback: a function which receives N arguments which correspond to each word recognized after the trigger word.

Example with variable arguments length

	trigger: ['write', 'escreva'],
	callback: function() {
		document.body.innerHTML = '<h1>' + [], ' ') + '</h1>';

Now, everything that you say after "write" (or in Portuguese "escreva") will be printed to the page.

Testing for browser support

The ultvoice object will be null in case the browser does not support the Speech Recognition API.

if (ultvoice) {
	//yay, do awesome speech recognition
} else {
	//browser does not support speech recognition, fallback or show error/warning message

Removing actions from ultvoice.actions

ultvoice.actions is a plain array. You can easily iterate over it, check for some criteria and remove items using .splice(), or assign to it a filtered copy of itself.


Set ultvoice.debug = true to get debugging info (such as every recognized speech input) printed to the console.

Native method wrappers

Upon abnormal interruption (other than ultvoice.stop()/.abort()), UltVoice will automatically ensue a request to restart speech recognition. You should always use the ultvoice.start()/.stop()/.abort() methods for this to work properly.

Exposed SpeechRecognition object

You can access UltVoice's SpeechRecognition object through ultvoice.listener. This way, you have access to the internal SpeechRecognition object and all of its API. It can be used for things that UltVoice does not abstract, e.g. setting its lang property.

AMD module

In case you'd like to use UltVoice as an AMD module, simply require it. In case UltVoice was require()d (e.g. through RequireJS) it will return the object instead of exposing a global variable.

    paths: {
        'ultvoice': 'path/to/ultvoice.min'

require(['ultvoice'], function(ultvoice) {
	//work with ultvoice as normal


Only Chrome has support for the Speech Recognition API thus far.


This library is currently in development, but I'll try to keep things stable in the master branch. Note that API function/property names may change so always check the changelog before updating.


  • trigger string with more than one word.

Feel free to open an issue if you have ideas for useful features, or submit a PR directly if you feel like. =]


  • 0.1: First revision.
You can’t perform that action at this time.