SpeechRTC

Speech recognition API built on top of webrtc using pocketsphinx to decode.

Demonstration: <br> http://www.youtube.com/watch?v=pnCRH-Iznrc <br> http://www.youtube.com/watch?v=cjjFvyH3kdc <br>

<h2>Client Implementation</h2>

    var speechrtc = new SpeechRTC("en-US");
    speechrtc.gram(["Apple","Oranges","Watermelon"]);
    speechrtc.listen();
    SpeechRTC.onRecognition = function(said)
    {
      console.log( " You said "+ said );
    }

<h2>Todo</h2>

• Dictation and continuous recognition (as this http://www.youtube.com/watch?v=3lTtCFaQF2A ) <br>
• Cleanup at onclose <br>
• Better demo page <br>
• Stabilize grammar swtich and jsgf write (remove garbage from socket) <br>
• Multi-lang support <br>
• 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! <br>

<h2>Running the server </h2> <h3> 1. Pre-requirements </h3> - libogg <br> - libopus <br> - pocketsphinx <br> - nodejs <br> - libpthread <br> - g++ <h3> Compiling & installing </h3> - Compile and install all dependecies above - Compile voiceserver <h3> Configuring </h3> - Configure and set the paths to models and audio storage on voiceserver <h3> Running </h3> - node server.js - ./voiceserver <h3> Acknowledgments </h3> - Steven Lee (Mozilla) - Shih-Chiang (SC) (Mozilla) - Robert O'Callahan (Mozilla) - Desigan Chinniah (Mozilla) - Randell Jesup (Mozilla) - All Mozilla Community <h3> Author </h3> Andre Natal<br> anatal@gmail.com