getusermedia-to-text

getUserMedia to Text via Google's Speech to Text API.

npm install getusermedia-to-text --save

Usage

var GetUserMediaToText = require('getusermedia-to-text')
var path = require('path')
var s2t = new GetUserMediaToText({
  projectId: 'your-project-id',
  keyFilename: path.join(__dirname, 'auth.json')
  // request: { options }
})
 
s2t.on('error', console.error)
s2t.on('status', console.log)
s2t.on('listening', function (isListening) {
  isListening ? console.log('Listening!') : console.log('Stopped Listening!')
})
s2t.on('data', console.log)
 
s2t.start()
s2t.stop()

Unfortunately google's speech API module depends on grpc which is a native module. Native modules don't work in the browser so this probably only works for electron after you run electron-rebuild.

To run the example:

npm install
npm start

You are required to set up your own service account since the speech recognition API requires an account with billing. See:

https://cloud.google.com/speech/docs/common/auth

The example code expects to find your service account credentials at ./example/auth.json.

API

var s2t = new GetUserMediaToText(options)

Create a new speech to text instance.

Options include:

• keyFilename (required): path to your auth.json file.
• projectId: your project id string. Default: getusermedia-to-text
• request: the request options object. Default:

{
  config: {
    encoding: 'LINEAR16',
    sampleRateHertz: 44100,
    languageCode: 'en-US'
  },
  singleUtterance: false,
  interimResults: false,
  verbose: true
}

s2t.start()

Start listening to audio. Will event speech events as data.

s2t.stop()

Stop listening to audio and destroy internal recognition stream.

s2t events:

• error: Emitted when an error is encountered with the error object.
• status: General status message strings.
• listening: Emits boolean listening status data when the listening status changes.
• data: Raw data objects that are emitted from the @google-cloud/speech library.