Live Streaming Audio Transcription

The transcription.live function encapsulates a websocket connection to the Deepgram API and returns a Node.js event emitter. Events received from the Deepgram API are emitted for consumption by your application.

Parameters

Additional options can be provided for streaming transcriptions. They are provided as an object of the transcription.live function. Each of these parameters map to a feature in the Deepgram API. Reference the
features documentation to learn what features may be appropriate for your request.

Initiating a Connection

The transcription.live function will return a Node.js event emitter that your application can subscribe to events on.

const deepgramLive = deepgram.transcription.live({
    punctuate: true,
    // additional options
});

Events

The following events are fired by the live transcription object:

Listening to Events

Use the addListener function to listen for events fired on the object returned from the transcription.live function.

deepgramLive.addListener("transcriptReceived", (transcription) => {
    console.log(transcription.data);
});

Functions

The object returned by the transcription.live function provides several functions to make using the Deepgram API easier. They are send, getReadyState, toggleNumerals, and finish.

Sending Data

The send function sends raw audio data to the Deepgram API.

deepgramLive.send(AUDIO_STREAM_DATA);

Get Ready State

The getReadyState function returns the ready state of the websocket connection to Deepgram.

const websocketReadyState = deepgramLive.getReadyState();

Toggle Features

The configure function gives the ability to toggle different features. Currently, only the numerals feature is supported with plans to add more in the future. The function takes in an object with features to toggle to determine whether or not to enable the feature.

deepgramLive.configure({numerals: true}); // enable numerals
deepgramLive.configure({numerals: false}); // disable numerals

This function is a shorthand way of sending the following data to the websocket connection:

deepgramLive.send(
    JSON.stringify({
        type: "Configure",
        processors: {
            numerals: true, // or false
        },
    })
);

Closing the Connection

The finish function closes the Websocket connection to Deepgram.

deepgramLive.finish();

Browser

Using the browser compatible SDK is slightly different. You still make the call to .live() and pass in the options you want applied to the transcription, but some of the events are different. We are abstracting the native WebSocket API.

import { Deepgram } from "@deepgram/sdk/browser";
const deepgram = new Deepgram("YOUR_API_KEY");
navigator.mediaDevices.getUserMedia({ audio: true }).then((stream) => {
    const mediaRecorder = new MediaRecorder(stream, {
        mimeType: "audio/webm",
    });
    const deepgramSocket = deepgram.transcription.live({
        punctuate: true,
    });
    deepgramSocket.addEventListener("open", () => {
        mediaRecorder.addEventListener("dataavailable", async (event) => {
            if (event.data.size > 0 && deepgramSocket.readyState == 1) {
                deepgramSocket.send(event.data);
            }
        });
        mediaRecorder.start(1000);
    });
    deepgramSocket.addEventListener("message", (message) => {
        const received = JSON.parse(message.data);
        const transcript = received.channel.alternatives[0].transcript;
        if (transcript && received.is_final) {
            console.log(transcript);
        }
    });
});