Getting Started With HTML5 Web Speech API

If you don’t know about it yet, the HTML5 web speech api specification is now in a working condition on google chrome and partially in apple safari browser(See the browser support status here: http://caniuse.com/web-speech). That means, you can now develop voice driven web applications. We can hope that other browsers will start supporting this very soon as well. In this tutorial, I will try to explain how we can start developing application that uses this and also refer you to a small wrapper library with easy to use abstraction that I wrote recently.

Voice To Text API:

We can alternatively mention it as ‘Speech Recognition API’ as well. What it does, is to capture user’s voice through input system and convert it to text. So, basically there is a need of voice recognition technology here. This feature is currently supported only in Google Chrome browser. By default, it uses google’s own voice recognition service. Here is a code example to implement it:

var recognizer = new webkitSpeechRecognition();
recognizer.lang = "en";
recognizer.onresult = function(event) {
    if (event.results.length > 0) {
        var result = event.results[event.results.length-1];
        if(result.isFinal) {
            console.log(result[0].transcript);
        }
    }  
};
recognizer.start();

The above code snippet will request permission from user to allow taking input through microphone access and then will capture the sound you talk, send it to external service for recognition, and get the result back inside ‘onresult’ event handler. Thus, we will be able to see the output in console window of browser.

This class definition also exposes an optional ‘serviceURI’ property which you can use to define the service url you like to use for voice recognition.

Text To Voice API:

Text to voice conversion is just a simple way to play a given text in robotic voice. Here is a simple code snippet for this:

var su = new SpeechSynthesisUtterance();
su.lang = "en";
su.text = "Hello World";
speechSynthesis.speak(su);

As you can see, its pretty much straight forward. Where we just need to pass the desired language and text to it and its all set to play it.

Checkout The HTML5 Web Speech Demo!

The Wrapper Library And Usage Example:

As I already experienced, some of the steps are obvious and can be made simple with a simple wrapper. Thus, to make your life easier, I started a small JavaScript library to ease the use of this api. Here is the github link:

https://github.com/ranacseruet/webspeech

It’s also registered as a bower package too! So, if you are using bower for front end package management, you can simply run command like:

$bower install webspeech

And you should be just fine!

Here is a very simple to use example:

<input id="text">
<button onclick="talk()">Talk It!</button>
<button onclick="listen()">Voice</button>
<script src="../bower_components/platform/platform.js"></script>
<script src="../src/webspeech.js"></script>
<script>
    var speaker = new webspeech.Speaker();
    var listener = new webspeech.Listener();
    function talk() {
        speaker.speak("en", document.getElementById("text").value);
    }

    function listen() {
        listener.listen("en", function(text) {
            document.getElementById("text").value = text;
        });
    }
</script>

Final Words:

Even on in chrome for windows, there is another issue, which is like it doesn’t support capturing voice continuously, instead, you will have to allow it on browser every time you want to say something. However, there is a work around to get rid of this annoying access allowance, which is to host your application on SSL. Then, only one time access will work for all later times.

Hope this simple tutorial will help you to get started with HTML5 web speech API easily without much difficulty. In case you are facing some issues, please let me know via commenting. Happy coding 🙂