This a fork of node-speakable that works with the WIT.AI API: https://wit.ai
Demo video on what you can achieve using node-speakable: https://my.doctape.com/s/NLHn1z
node-speakable is a continuous speech recognition module for node.js.
Basically, node-speakable is continuous waiting for you to say something and waits until you finally stopped talking. It then emits an speechResult
event with an Object()
including your Intents. You can then parse the response to trigger some awesome action to happen, like turning on your Philips Hue
lights.
If you ever talked to your XBOX360 (kinect) you're already familiar on how a continuous speech recognition system works for you.
It's pure JavaScript magic... Ok, not yet! Currently node-speakable needs you to put a binary of sox
into the modules (into lib) folder to do the recording. The actual voice recognition is then achieved trough a POST to the the Google Speech API.
You'll need to ensure that you turn the Google Speech API selection to 'ON' in the API Console. If you don't see it there, you might want to check out this first: http://www.chromium.org/developers/how-tos/api-keys. For the example project, you'll also need to set the environment variable GKEY in your shell to your API key.
var Speakable = require('./');
var speakable = new Speakable({key: 'your-WIT-API-key'});
speakable.on('speechStart', function() {
console.log('onSpeechStart');
});
speakable.on('speechStop', function() {
console.log('onSpeechStop');
});
speakable.on('speechReady', function() {
console.log('onSpeechReady');
});
speakable.on('error', function(err) {
console.log('onError:');
console.log(err);
speakable.recordVoice();
});
speakable.on('speechResult', function(recognizedWords) {
console.log('onSpeechResult:')
console.log(recognizedWords);
speakable.recordVoice();
});
speakable.recordVoice();
node-speakable is licensed under the MIT license.
- Allow overwrite of speakable’s defaults (sox path & parameters, etc.)
- Limit maximum listening time