greatcall-firefox / node-wit Goto Github PK

1. You need to create an Wit instance first.

2. Install Node.JS on your computer.

3. Setup your project

   Create a new Node.JS app :

   $ mkdir myapp
   $ cd myapp
   $ npm init
   ...

   Add node-wit as a dependencies in your package.json

     "dependencies": {
       "node-wit": "1.1.0"	
     },	

   Execute npm install in your current folder to fetch the dependencies

   We will send an audio file to Wit.AI

   You can use SoX to record WAV files from the command line. brew install sox on OSX and apt-get install sox on ubuntu. The following options will create a Wit-ready WAV file (press Ctrl+C to stop recording):

   sox -d -b 16 -c 1 -r 16k sample.wav

   Create a index.js file in myapp directory containing:

   var wit = require('node-wit');
   var fs = require('fs');
   var ACCESS_TOKEN = "IQ77NWUPUMNBYEUEKRTWU3VDR5YSLHTA";
   
   console.log("Sending text & audio to Wit.AI");
   
   wit.captureTextIntent(ACCESS_TOKEN, "Hello world", function (err, res) {
       console.log("Response from Wit for text input: ");
       if (err) console.log("Error: ", err);
       console.log(JSON.stringify(res, null, " "));
   });
   
   var stream = fs.createReadStream('sample.wav');
   wit.captureSpeechIntent(ACCESS_TOKEN, stream, "audio/wav", function (err, res) {
       console.log("Response from Wit for audio stream: ");
       if (err) console.log("Error: ", err);
       console.log(JSON.stringify(res, null, " "));
   });

4. Start your app

$ node index.js
Sending text & audio to Wit.AI
Response from Wit for text input:
{
 "msg_id": "b46d4a08-1e2e-43f4-b30a-aaa7bccb88e3",
 "_text": "Hello world",
 "outcomes": [
  {
   "_text": "Hello world",
   "intent": "greetings_hi",
   "entities": {},
   "confidence": 0.929
  }
 ]
}
Response from Wit for audio stream:
{
 "msg_id": "83c14e47-13cb-4ad4-9f5e-723cd47016be",
 "_text": "what's the weather in New York",
 "outcomes": [
  {
   "_text": "what's the weather in New York",
   "intent": "weather",
   "entities": {
    "location": [
     {
      "suggested": true,
      "value": "New York"
     }
    ]
   },
   "confidence": 1
  }
 ]
}

The captureTextIntent function returns the meaning extracted from the text input. The function takes 3 parameters:

• access_token: Your access token for your instance
• text: The text input you want to extract the meaning of
• callback(error, response): A callback function get 2 arguments:
  1. An error when applicable
  2. A JSON object containing the Wit.AI response

var wit = require('node-wit');
wit.captureTextIntent(ACCESS_TOKEN, "Hello world", function (err, res) {
    console.log("Response from Wit for text input: ");
    if (err) console.log("Error: ", err);
    console.log(JSON.stringify(res, null, " "));
});

The captureSpeechIntent function returns the meaning extracted from the audio input. The function takes 4 arguments:

• access_token: Your access token for your instance
• stream: The audio stream you want to extract the meaning of
• content-type: The content-type of your audio stream (audio/wav, audio/mpeg3, audio/raw;encoding=unsigned-integer;bits=16;rate=8000;endian=big, ...)
• callback(error, response): A callback function get 2 arguments:
  1. An error when applicable
  2. A JSON object containing the Wit.AI response

var wit = require('node-wit');
var fs = require('fs');
var stream = fs.createReadStream('sample.wav');
wit.captureSpeechIntent(ACCESS_TOKEN, stream, "audio/wav", function (err, res) {
    console.log("Response from Wit for audio stream: ");
    if (err) console.log("Error: ", err);
    console.log(JSON.stringify(res, null, " "));
});