Complete code and the description is on our github site: https://github.com/voicegain/platform/tree/master/examples/offline-transcription-demo

Requires Voicegain Platform Release 1.9.0 or higher

Intro

What this code demonstrates:

• hiding JWT on the server side
• upload of file from browser to Voicegain storage
• submit it to offline transcription
• poll until transcription is finished
• show the text result of transcription

Steps to Install and Run

1. In the Voicegain Web Console add localhost:8081 to allowed domains (Settings -> API Security) - this is needed to prevent CORS rejects in the browser
2. Install node js - https://nodejs.org/en/download/
3. Set JWT in server.js
4. Run command npm install in project directory to install dependencies
5. To start server, run command node server.js while in project directory
6. In a different terminal window start http-server
7. In a web browser goto http://localhost:8081/index.html (we are assuming that http-server will run on 8081 port as node.js is using 8080)

Notes about the code

server.js

Encapsulates Voicegain API to generate short-lived JWT tokens that are needed to make all the other Voicegain Web API requests.

This file contains your Voicegain account JWT token (in <Your JWT Here>) that gives you access to all the Voicegain APIs. It is important to hide it on the server side, otherwise, someone could get easy access to it and run up API requests on your account and you would have to pay for someone else's usage.

In the server.js included in this demo we set the short-lived JWT token parameters to:

• aud: "localhost:8081" - this is the authorized domain for this token
• expInSec: 3600 - this is the expiry time of the short lived token - setting it to 1 hour is a bit extreme, so we suggest using smaller values in prod applications.

upload.js

This is the main javascript for the application. It contains:

• Fetching of the short-lived JWT from node.js server http://localhost:8080/api/jwt
• Audio file upload using POST https://api.voicegain.ai/v1/data/file API
• Start of asynchronous transcription of the uploaded file using POST https://api.voicegain.ai/v1/asr/transcribe/async API
• Periodic polling of the status of transcription using GET https://api.voicegain.ai/v1/asr/transcribe/${sessionId} API
  • note that we set full=true in the polling query as we are not interested in the incremental transcription results - just the final one
• When result.final==true, the retrieval of the text transcript from the polling response.

Possible extensions

Voicegain API provides GET https://api.voicegain.ai/v1//asr/transcribe/{sessionId}/transcript API that allows download of the final transcript in a variety of formats:

• json - complete transcript data with all detail for each word
• text - plaint text transcript with optional timing information
• srt - transcript in SRT caption format
• vtt - transcript in WebVTT caption format
• ttml - transcript in TTML (XML) caption format