Traditional server applications typically need to be always on, always using resources and require maintenance to ensure availability. Serverless works differently - functionality is exposed via URLs. When a request is made they spin up, execute logic, and spin back down. Each serverless script (known as a 'function') can be run many times in parallel, so this approach may be suitable for scale, dependent on your use case.

In this tutorial, you will set up serverless functions with Netlify to get transcripts using Deepgram's Node.js SDK - one for hosted files, and one for local files. Because serverless functions do not lend themselves to long-living scripts, it's not recommended to use this approach for live transcription.

Before We Start

You will need:

Create a new directory and navigate to it with your terminal. Run npm init -y to create a package.json file and then install the Deepgram Node.js SDK:

npm install @deepgram/sdk

Set Up Netlify Project

You can set up a Netlify project from the web dashboard, but as we need the Netlify CLI to test our functions, we may as well use it here. Inside of your project directory, run netlify init, and when prompted, choose Create and deploy site manually.

A new project will now be visible in your Netlify web dashboard - you can open it with netlify open.

Your First Netlify Function

Netlify offer zero-configuration serverless functions if you put your logic in a specific directory - /netlify/functions. Create a new file at /netlify/functions/hello.js and populate it with the following:

exports.handler = async (event) => {
  try {
    // Any logic goes here, but we'll return a fixed response
    return { statusCode: 200, body: JSON.stringify({ message: 'ok' }) }
  } catch (err) {
    return { statusCode: 500, body: String(err) }

Test Your Function

Run netlify dev and wait for the local server to start - usually at http://localhost:8888. Open another terminal and run the following command to see the response:

curl http://localhost:8888/.netlify/functions/hello

<Alert type="info">Don't be alarmed by the . in the URL - your local directory, which contains your functions, should just be /netlify</Alert>

Your terminal should look something like this:

Adding Your Deepgram API Key

Like most hosting providers, Netlify provides a way to set sensitive keys as environment variables. Netlify CLI will inject any variables from your web dashboard to your local runtime for you - super cool.

Open your project dashboard with netlify open while in your project directory. Heard to Site settings > Build & deploy > Environment > Environment variables and create a new variable called DEEPGRAM_API_KEY with the value from the Deepgram Console.

If you are still running your netlify dev server, stop it with ctrl + c and restart it. You should see the key being injected,meaning it is now available with process.env.DEEPGRAM_API_KEY

Transcribe Hosted Files

Inside of your functions directory, create hosted.js with the following content:

const { Deepgram } = require('@deepgram/sdk')
const deepgram = new Deepgram(process.env.DEEPGRAM_API_KEY)

exports.handler = async (event) => {
  try {
    const { url } = JSON.parse(event.body)
    const { results } = await deepgram.transcription.preRecorded({ url })
    return { statusCode: 200, body: JSON.stringify(results) }
  } catch (err) {
    return { statusCode: 500, body: String(err) }

Once you save the file, the new URL is immediately available. This function requires a data payload with a url property. You can test it by once again using cURL:

curl -X POST -H "Content-Type: application/json" -d '{"url": ""}' http://localhost:8888/.netlify/functions/hosted

Accessing Functions From The Web

Netlify makes your functions available on the same domain as your main application (just under the /.netlify/functions path). Due to this, we can call Netlify Functions from our main application by specifying the relative URL. This means it will work both locally and once deployed.

Create an index.html file in your main directory:

<!DOCTYPE html>
    <meta charset="UTF-8" />
    <button>Transcribe from URL</button>
      document.querySelector('button').addEventListener('click', () => {
        const url = prompt('Please provide an audio file URL')
        fetch('/.netlify/functions/hosted', {
          method: 'POST',
          body: JSON.stringify({ url }),
          .then((r) => r.json())
          .then((data) => {

Navigate to http://localhost:8888 in your browser, click the button, and provide a static file URL (if you don't have one, use Open your browser console, and you should see the response from Deepgram.

Transcribe Local Files

Create a new functions file - file.js:

const { Deepgram } = require('@deepgram/sdk')
const deepgram = new Deepgram(process.env.DEEPGRAM_API_KEY)

exports.handler = async (event) => {
  try {
    const { results } = await deepgram.transcription.preRecorded({
      buffer: Buffer.from(event.body, 'base64'),
      mimetype: 'audio/wav',
    return { statusCode: 200, body: JSON.stringify(results) }
  } catch (err) {
    return { statusCode: 500, body: String(err) }

Add a <form> just below the <button> in index.html:

  <input id="file" type="file" name="file" />
  <input type="submit" value="POST to server" />

Refresh your browser and upload a file - you should see the results in your browser. If you want to handle the results within the page, Sandra details how to submit a form using JavaScript here.

Deploying Functions

Ready? netlify deploy. That's it.

Once deployed, you'll be able to access your Netlify functions at Your webpage will work without modifications because it will be served at the same subdomain.

Netlify also supports push-to-deploy with GitHub if you configure your project to deploy from a repo.

Wrapping Up

Netlify makes deploying serverless functions reasonably straightforward, but if you have any questions after reading this guide, we are here to help! Just pop us a line at @Deepgramai.

If you have any feedback about this post, or anything else around Deepgram, we'd love to hear from you. Please let us know in our GitHub discussions .

Related Articles

Unlock language AI at scale with an API call.

Get conversational intelligence with transcription and understanding on the world's best speech AI platform.

Sign Up FreeBook a Demo
Essential Building Blocks for Language AI