@gkt/transcriber NPM

@gkt/transcriber

@gkt/transcriber is a JavaScript library for performing voice transcriptions via GreenKey's backend services.

Installation

$ npm install @gkt/transcriber

Or via unpkg.com:

<script type="text/javascript" src="https://unpkg.com/@gkt/transcriber"></script>

Usage

// es6 module
import Transcriber from '@gkt/transcriber';

// common js module
// const Transcriber = require('@gkt/transcriber');

// script tag
// const Transcriber = window.Transcriber;

// Provide a base URL that is proxied to your GreenKey backend instance
const transcriber = new Transcriber({gkUrl: '/greenkey'});

// This will prompt the user for access to the microphone
transcriber.init().then(function() {
  // Register a callback for transcription data. Updates will be emitted periodically while a
  // transcription is active. A final data object will be emitted after the transcription is stopped
  // and is indicated by the `data.final` property.
  transcriber.on(Transcriber.TRANSCRIBER_DATA_RECEIVED, function(t, data) {
    console.log(data);
  });

  // Start recording audio data from the microphone and send it to the GreenKey backend for
  // processing.
  transcriber.start();

  // Stop recording audio data.
  transcriber.stop();
});

Processing Results

Transcriptions are produced in real time and communicated to the client via the Transcriber.TRANSCRIBER_DATA_RECEIVED event. The first argument to the event handler is the transcriber instance and the second is an object containing the current result of the transcription. The following top-level fields are included:

Field	Type	Description
session	string	A unique identifier for the transcription session
final	boolean	Indicates whether this is the final version of the transcript
clockStart	string (ISO8601 date)	The time the transcription recording started
segments	array of objects	A list of segment objects

And the segments array contains objects with these fields:

Field	Type	Description
final	boolean	Indicates whether this segment is finished processing
startTimeSec	number	The offset in seconds from `clockStart` where the segment audio started
endTimeSec	number	The offset in seconds from `clockStart` where the segment audio ended
transcript	string	The text of the raw transcript
interpreted_quote	object	Present when a quote is detected in the transcript
interpreted_quote.imString	string	The generated IM String of the detected quote
interpreted_quote.productClass.label	string	The product class of the detected quote
words	array	A list of transcript works with confidence levels

Example Apps

Some example apps are provided in the examples directory. Edit the .html files to set the gkUrl variable to your GreenKey instance before opening in a browser.

The examples are full apps that send audio data recorded from the microphone, receive transcription results and display them.

Notes

This library uses the MediaDevices.getUserMedia() API for gaining access to the user's microphone hardware. This API is only available in secure contexts, so your application must be served over HTTPS in order for it to work.