Speech recognition API built on top of webrtc using pocketsphinx to decode.
Demonstration:
http://www.youtube.com/watch?v=pnCRH-Iznrc
http://www.youtube.com/watch?v=cjjFvyH3kdc
var speechrtc = new SpeechRTC("en-US");
speechrtc.gram(["Apple","Oranges","Watermelon"]);
speechrtc.listen();
SpeechRTC.onRecognition = function(said)
{
console.log( " You said "+ said );
}
- Dictation and continuous recognition (as this http://www.youtube.com/watch?v=3lTtCFaQF2A )
- Cleanup at onclose
- Better demo page
- Stabilize grammar swtich and jsgf write (remove garbage from socket)
- Multi-lang support
- Connect Web Speech API at Firefox (https://dvcs.w3.org/hg/speech-api/raw-file/tip/speechapi.html and https://bugzilla.mozilla.org/show_bug.cgi?id=650295)
- Improve the installation procedure and documentation!
- libopus
- pocketsphinx
- nodejs
- libpthread
- g++ - Compile and install all dependecies above - Compile voiceserver - Configure and set the paths to models and audio storage on voiceserver - node server.js - ./voiceserver - Steven Lee (Mozilla) - Shih-Chiang (SC) (Mozilla) - Robert O'Callahan (Mozilla) - Desigan Chinniah (Mozilla) - Randell Jesup (Mozilla) - All Mozilla Community Andre Natal
anatal@gmail.com