AudioCodes (LiveHub) and Deepgram STT

AudioCodes VoiceAI Connect is a powerful platform that enables the integration of telephony and contact center platforms with the cloud, thus facilitating the use of Deepgram in your customer journey.

It offers a simple user interface for standing up connections and developer-friendly APIs for advanced integrations.

Note, AudioCodes offers two versions of this platform - LiveHub, which is the self serve version, as well as VoiceAI Connect Enterprise, which is the managed services version.

This guide will focus on LiveHub, as it is accessible to all users. However, if you are using VoiceAI Connect Enterprise, the steps and instructions laid out here will be very similar. In addition, you will have the help of the AudioCodes Professional Services team to help you with the configuration. Refer to this doc for guidance, but note that the steps may vary slightly depending on how your specific version of VoiceAI Connect Enterprise is built.

Before you Begin

Add Deepgram as a Speech Service in LiveHub

Follow the steps in the LiveHub integration guide to create a new speech service.

If you want to use Deepgram’s Speech-to-Text or Text-to-Speech services, you should choose the Deepgram specific speech provider option.

You can also use the Custom Integration option and insert the Deepgram API endpoint (example - wss://api.deepgram.com/v1/listen).

Using the Deepgram specific speech service

Selecting this option automaticaly routes your requests to the default Deepgram STT or TTS endpoints:

• Speech-to-Text api.deepgram.com/v1/listen
• Text-to-Speech api.deepgram.com/v1/speak

Using the Custom Integration option

Select this option if you wish to pass the Deepgram API endpoint explicitly, or for testing another alternate endpoint.

Advanced Configuration JSON object

AudioCodes supports Advanced JSON configuration in the Bot Connection. This is how you can pass additional query parameters to Deepgram.

Follow these instructions if you need to set up a Bot Connection for this first time.

To edit an existing Bot Connection, navigate to Bot Connections > Edit > Advanced

Please see the AudioCodes documentation for the full list of supported key-value pairs in the Advanced JSON configuration.

Deepgram officially supports the following AudioCodes configuration parameters:

• sttLanguage - use this to set the language code for STT. This is the ONLY way to change the language!
• Other / Misc parameters - the following parameters are suppoted by Deepgram. You can set any of these to your desired values. NOTE: requires VoiceAI Connect Enterprise (Version 3.22 and later).
• sttGenericData - you can use this object to pass ANY parameter (except language) to Deepgram.

Using the sttGenericData object

If you wish to pass ANY supported parameter (other than language) to Deepgram, you can do so using the sttGenericData object. For example, any of the parameters in our API documentation can be passed through the sttGenericDataobject.

For example, you might configure the sttGenericData object as such:

1
{
2
  "sttGenericData": {
3
    "model": "nova-3", // we will default to 'phonecall-enhanced' if not specified here
4
    "smart_format": true, // str or bool supported
5
    "profanity_filter": "true" // str or bool supported
6
  }
7
}

Using Keyterm or Keyword boosting

If you wish to use Keyterms (nova-3) or Keywords (nova-2, nova-1, base, enhanced), you should pass a list through the sttGenericData object. See the example below.

Keyword Intensifiers are supported. Simply pass the value immediately after the keyword, separated by a colon, as per our documentation.

1
{
2
  "sttGenericData": {
3
    // "model": "nova-3" OR "nova-2"
4
    "keyterm": ["example", "another example"] // For nova-3
5
    "keywords": ["snuffleupagus:5"] // For nova-2, nova-1, base, or enhanced models
6
  }
7
}

Accessing Deepgram AI Works flows

1. Configure the Custom Integration Speech Service

   • Set up a Custom Integration speech service pointing to the Deepgram AudioCodes integration endpoint
   • Use the endpoint: wss://integrations.deepgram.com/audiocodes/stt
   • Configure the sttGenericData object to pass your parameters

2. Define and Pass the AI Works Object

   • Define an aiworks object that routes requests to AI Works (instead of the default API endpoint)
   • Include the AI Works endpoint that Deepgram has provided to you
   • Pass any additional query parameters required for your integration

1
{
2
  "sttGenericData": {
3
    "aiworks": {
4
      "repo": "wss://aiworks.deepgram.com/api/repos/<example-repo-id>/flows/<example-flow-id>/execute", // Insert your AI Works URL here!
5
      "stt_query_params": { // If using AI Works, you must set the STT query parameters here, NOT at the top level. 
6
        "model": "nova-3",
7
        "smart_format": true, // str or bool supported
8
        "mip_opt_out": "true" // str or bool supported
9
      }
10
    }
11
  }
12
}