yandex-recognizer

Node JS interface to yandex speech recognition service

Questions and Bug Reports

• mailing list: [email protected]

Installation

Install the yandex-recognizer and it's dependencies by executing the following NPM command.

npm install yandex-recognizer --save

Troubleshooting

The yandex-recognizer depends on several other packages. These are.

• node-uuid
• websocket

Ensure, that your user has write permission to wherever the node modules are being installed.

QuickStart

Add require to your app:

var yandex = require('yandex-recognizer');

Create service:

var service = yandex.Recognizer(config);

config object must define at least 5 properties:

config = {

	onConnect: function(sessionId, code),
	onResult: function(data),
	onError: function(e),
	onClose: function(e),
	apikey: 'YOUR-OWN-YANDEX-API-KEY'

}

// All other properties will be set by default (see source code for all available).

Connect service to your app:

service.connect();

onConnect(sessionId, code) method will launched in case of successfull connect. Here you send sound data for recognition:

service.send(db, cbLength);

// db (typeof Buffer) - your sound file data.
// cbLength (typeof Number) - length of chunk buffer.

onResult(data) method receive recognized text and some other data.

onError(e) method receive error massages.

onClose(e) method receive close massage.

Limitations

Input audio file format may be now only:

• Sample Encoding: 16-bit Signed Integer PCM, Sample Rate: 16000, 1 Channel.

Next Steps

• yandex-recognizer documentation
• node example
• ajax cgi