ringcentral-softphone

RingCentral Softphone SDK for JavaScript

Experimental

This project is still in its early stage. It is not ready for production yet!

What are the differences between ringcentral-web-phone and this project?

ringcentral-web-phone is designed for client side and only works with browsers.

This project was originally designed for server and desktop. It works both with and without browsers.

Install

yarn add ringcentral-softphone @ringcentral/sdk

For node.js you also need to:

yarn add ws wrtc

because node.js by default doesn't support WebSocket & WebRTC.

Usage

import RingCentral from '@ringcentral/sdk'
import Softphone from 'ringcentral-softphpone'

const rc = new RingCentral(...)
await rc.login(...)
const softphone = new Softphone(rc)
await softphone.register()
// to be continued

Demos

Setup

yarn install
cp .env.sample .env

Edit .env file to specify credentials.

Note: please do NOT get rid of NODE_TLS_REJECT_UNAUTHORIZED=0 in .env file. It is required for WebSocket to work without a TLS certificate installed on your laptop. You may remove it however if your app target browsers instead of node.js.

Run

For node.js:

yarn server

For browsers:

yarn browser

Test

Make a phone call to the phone number you configured in .env file.

for node.js

The app will auto pick up the call and save your voice to audio.raw

You can play audio.raw file by command line

play -b 16 -e signed -c 1 -r 48000 audio.raw

If play command is not available, please install sox: brew install sox.

for browsers

The app will auto pick up the call and redirect your voice to an <audio/> HTML5 element.

Interesting Usage cases

Call supervision

Let's say there is a phone call ongoing between a customer and the call agent.

You can use this library to supervise this phone call to get live audio stream.

You can analyze the audio stream using some AI algorithm and provide tips to the call agent in real time.

Live transcription

Use this library to supervise an existing phone call to get live audio stream.

Translate the audio stream into text by invoking some speech-to-text service.

Show the text to the caller and/or callee so they can see live transcription.

Todo

• state machine
• make outbound call
• release to NPM