JavaScript Speech Recognition
Elias sent us a link to his JavaScript speech recognition API at speechapi.com. It uses Flash to record audio and sends it to a server, then the server returns suggestions to a callback. You have to register to use the API. Using the API looks like this:
<script type="text/javascript" src="http://www.speechapi.com/speechle.js"></script>
<script language="JavaScript" type="text/javascript">
var speechle;
function TheSettings() {
var info = {'Userid':'eli', 'Logging':'true'};
var eventCallbacks={'OnResult':'result', 'OnLogging':'logging', 'OnFinishTTS':'finishTTS'};
var grammar={'Text':'one,two', 'Type':'simple'};
speechle = new SPEECHLE(info,eventCallbacks,grammar);
}
The site has a few demos. One which worked pretty reliably for me was the random quiz.
A similar technology is WAMI which uses Java instead. You can try a demo in the WAMI sandbox.
Standards
There are W3C standards relating to speech recognition:
- Speech Recognition Grammar Specification
- Voice Extensible Markup Language — a modular XML language for creating interactive media dialogs that feature synthesized speech, recognition of spoken and DTMF key input
See the Voice Browser Activity for a centralised pool of related W3C resources.