Genesys and Deepgram

In this guide, we'll explain how to configure Genesys to share call audio will Deepgram and have transcriptions sent to a callback URL of your choice

Before you Begin

Create a Deepgram Account

Before you can use Deepgram products, you'll need to create a Deepgram account. Signup is free and includes:

• $200 in credit, which gives you access to:
  • All public models
  • Pre-recorded and Live Streaming features
  • Audio Intelligence features
  • Text to Speech features
  • Text Intelligence features

• all features

Create a Deepgram API Key

To access Deepgram’s API, you'll need to create a Deepgram API Key. Make note of your API Key; you will need it later.

Set Up a Genesys Cloud Org

You'll also need a Genesys Cloud org where you have completed the initial setup tasks so that your call center can receive calls.

Install AudioHook Monitor

In your Genesys org, install AudioHook Monitor. Note that you will also need to configure voice transcription in order for AudioHook Monitor to work.

Configure AudioHook Monitor

Once you've installed AudioHook Monitor, it will appear under Admin > Integrations in the Genesys UI, as in the image below. Now click the three dots on the AudioHook row and choose Edit Integration.

You should now see an interface like below. Under the Details tab, you can give your AudioHook instance a more useful name. Let's name it Deepgram.

Under Configuration > Properties, set Channel to both and Connection URI to wss://integrations.deepgram.com/genesys.

Under Configuration > Credentials, click Configure and paste your Deepgram API key into the API Key field. Leave the Client Secret field blank. Then click OK.

📘

If you specify an API key and then change it, it can take a long time for the new API key to propagate through the Genesys system. Give it 30 minutes before you assume that Deepgram is receiving the new API key. If you don't want to wait, you can delete AudioHook from the Integrations page, reinstall it, and provide the new API key to the reinstalled AudioHook.

Under Configuration > Advanced, provide a JSON object to customize the Deepgram request. For example:

{
  "model": "general",
  "tier": "nova",
  "punctuate": true,
  "smart_format": true,
  "tag": [
    "sometag1",
    "sometag2"
  ],
  "callback": "https://example.com/webhook/{conversation-id}"
}

Within this JSON object, you have access to the full suite of features in Deepgram's streaming API. However, be careful NOT to include any of these:

• sample_rate
• encoding
• channels
• multichannel

Deepgram already knows the sample_rate, encoding, and channels based on the information sent from Genesys, and multichannel is locked to true so that you can tell who is speaking in your transcripts. If you include any of these four features, the integration will fail!

Callback

Let's take a moment to look at the most important feature: the callback. The callback is the URL where your transcripts will be POSTed in real-time. It works exactly like it does for a normal streaming request, except when you are using Deepgram with Genesys, there are three special placeholders you can include in the callback URL:

• {conversation-id}
• {participant-id}
• {session-id}

In the example above, we provide a callback of https://example.com/webhook/{conversation-id}. This means that if there is a Genesys call with conversation id fb486691-fbe1-4b8d-a28e-72ca891e8f0b, the transcripts for that call will be POSTed to https://example.com/webhook/fb486691-fbe1-4b8d-a28e-72ca891e8f0b. This allows us to match up transcripts with calls.

To quickly validate that everything is working, it can be helpful to point your callback URL to a free tool that displays incoming requests. Consider using beeceptor.com, webhook.site, or--with slightly more effort but less rate limiting--ngrok.

Understand Which Calls Will Be Transcribed

Now we are done with configuration and can click Save. Before flipping the switch to activate the integration, we should understand which calls will be transcribed by Deepgram. This information is available in the AudioHook docs, but it bears repeating here because it's easy to use far more Deepgram credit than you intended if you're not careful.

For historical reasons, Genesys built AudioHook on top of its transcription feature. This means AudioHook integrations will be used wherever transcription is active. At the time of writing, this can happen in two ways:

1. Transcription is active at the level of a queue, or
2. Transcription gets enabled with a Transcription action in an architect flow.

To check your queues, go to Admin > Contact Center > Queues. For any queue where Voice > Voice Transcription is set to On, all calls going through that queue will be transcribed by Deepgram and have their transcripts sent to the callback URL you specified.

To check your architect flows, go through each flow and look for one of these blocks:

This block will trigger the Deepgram integration unless you've suppressed recording for the flow under Settings > Recording and Speech Recognition.

Once you've checked that Deepgram will be used exactly where you want, it's time to activate the integration.

Activate Integration

Back in Admin > Integrations, flip the switch to activate the integration. Make sure it goes into the Active state and no errors are displayed, like the image below.

At this point, if everything was done correctly, you should be able to put a call through the system and see your transcripts arriving at your specified callback URL.