Get Live Transcription - Phone Call (Node.js) Telephony

Get the live transcription in your Node.js application by making a call to a valid phone number.

Making a phone call is also the quickest way to test Symbl’s Telephony API. It can make an outbound call to a phone number using a traditional public switched telephony network (PSTN), any SIP trunks, or SIP endpoints that can be accessed over the internet using a SIP URI.

Set up Symbl#

To get started, you’ll need your account credentials and Node.js installed (> v8.x) on your machine.

We’ll use the Symbl module for Node.js in this guide. Make sure you have a Node project set up. If you don’t have one, you can set one up using npm init.

From the root directory of your project, run the following command to add symbl-node in your project dependencies.

$ npm i --save symbl-node

Retrieve your credentials#

Your credentials include your appId and appSecret. You can find them on the home page of the platform. credentials page

Initialize SDK#

const {sdk, SpeakerEvent} = require("symbl-node");

sdk.init({
    // Your appId and appSecret https://platform.symbl.ai
    appId: 'your_appId',
    appSecret: 'your_appSecret'
}).then(async () => {
    console.log('SDK initialized.');
    try {
      // You code goes here.
    } catch (e) {
        console.error(e);
    }
}).catch(err => console.error('Error in SDK initialization.', err));

In a new file called index.js in the project, add the following lines to import and initialize the SDK. Replace the appId and appSecret in the code with yours.

Make a call#

The quickest way to test the Telephony API is to make a call to any valid phone number. The Telephony API only works with phone numbers in the U.S. and Canada.

tutorial phone integration

Your application uses the SDK to:
- make REST call to Telephony API with phone number details and then Symbl makes the call
- subscribe to live results over a WebSocket connection
Once the call is accepted on the phone, the Telephony API starts receiving audio from the call.
Live transcription is streamed back to your application using the subscribed WebSocket connection.

SIP
PSTN

const connection = await sdk.startEndpoint({
    endpoint: {
        type: 'sip', // when dialing in to SIP UAS or making a call to SIP Phone
        // Replace this with a real SIP URI
        uri: 'sip:bob@sip.example.com'
    }
});
const {connectionId} = connection;
console.log('Successfully connected. Connection Id: ', connectionId);

To make a phone call, call the startEndpoint with type set to ‘pstn’ and a valid U.S./Canadian phone number phoneNumber.

Subscribe to the Live Results#

To get the live transcription, you need to subscribe to the connection to get the results. You need to call the subscribeToConnection method in the SDK and pass the connectionId and a callback method which will be called on for every new event including the (live) transcription.

// Subscribe to connection using connectionId.
sdk.subscribeToConnection(connectionId, (data) => {
  const {type} = data;
  if (type === 'transcript_response') {
      const {payload} = data;

      // You get live transcription here!!
      process.stdout.write('Live: ' + payload && payload.content + '\r');

  } else if (type === 'message_response') {
      const {messages} = data;

      // You get processed messages in the transcript here!!! Real-time but not live! :)
      messages.forEach(message => {
          process.stdout.write('Message: ' + message.payload.content + '\n');
      });
  } else if (type === 'insight_response') {
      const {insights} = data;
      // You get any insights here!!!
      insights.forEach(insight => {
          process.stdout.write(`Insight: ${insight.type} - ${insight.text} \n\n`);
      });
  }
});

When you use this API, you can use a callback function within the subscribeToConnection function. You pass an object and declare the type of response you want in the type field.

transcript_response type is used for low latency live transcription results
message_response type is used for processed transcription results in real-time. These are not the same as transcript_response which are low latency but typically generate a few seconds after the processing and contain messages split by sentence boundaries.
insight_response is used for any insights detected in real-time

End the Call#

To end the call, you should make a stopEndpoint call. The following code stops the call after 60 seconds. Your business logic should determine when the call should end.

// Stop call after 60 seconds to automatically.
setTimeout(async () => {
  const connection = await sdk.stopEndpoint({ connectionId });
  console.log('Stopped the connection');
  console.log('Conversation ID:', connection.conversationId);
}, 60000); // Change the 60000 with higher value if you want this to continue for more time.

The stopEndpoint will return an updated connection object which will have the conversationId in the response. You can use conversationId to fetch the results even after the call using the Conversation API.

Code#

You can find the complete code used in this guide here.

Test#

To verify and check if the code is working:

Run your code:

$ node index.js

You should receive a phone call to the number you used in the startEndpoint call. Accept the call.
Start speaking in English (default language) and you should see the live transcription added to the console in real-time.
The call should automatically end after 60 seconds. If you end it sooner and don’t invoke stopEndpoint, you will not receive the conversationId. If you need to access the results generated in the call, you should invoke stopEndpoint even if it was ended without explicitly invoking stopEndpoint prior to this point.

Note#

The example above invokes stopEndpoint after a fixed timeout of 60 seconds. This is for demonstrating the stop functionality and it is not the recommended method of implementation for your application. In a real implementation, you should invoke startEndpoint and stopEndpoint as needed by the business logic of your application, i.e when you would like Symbl to start processing and stop processing.