Skip to content
Nappai Documentation

Deepgram STT

Deepgram STT is one of the components that act as the hearing of a voice agent: it turns what the caller says into text so the agent can understand it. It uses Deepgram’s speech recognition service and transcribes while the person is still speaking, so the agent can start preparing its answer before the sentence ends.

It plugs into the STT connector of the Voice Agent node. It holds the speech recognition settings the agent uses on every call.

When to use it

  • You want cloud speech recognition for an agent that takes calls in Spanish or English.
  • Your callers often interrupt the agent and you want the word-based interruption settings of the Voice Agent to work.
  • You already have a Deepgram account and want to use the same one for the agent’s voice with Deepgram TTS.
  • You want to pick the Deepgram model that transcribes, among those your account offers for the call’s language.

Configuration

You need a Deepgram API credential with your Deepgram account key. See Credentials.

FieldWhat it does
Credential (required)The Deepgram API credential used to connect to Deepgram.
LanguageCode of the language the caller is expected to speak, for example es or en. Default: es.
ModelThe Deepgram model that transcribes. Default: nova-3. Once you pick the credential, the list shows the Deepgram models that transcribe the chosen language.

Output

The STT output holds the speech recognition settings. It connects to the STT connector of the Voice Agent node, or to Fallback STT when you turn on Fallback Engines and want it as a backup. The editor will not let you plug it into a TTS connector by mistake.

Example

A clinic takes appointment calls in Spanish with a voice agent.

  1. Add Deepgram STT to the flow and connect it to the STT connector of the Voice Agent.
  2. In Credential, choose the clinic’s Deepgram API credential.
  3. Leave Language as es and Model as nova-3.
  4. Test the agent in the playground by asking for an appointment the way a patient would.

Good practices

  • Deepgram charges per minute of audio on your Deepgram account. Check your usage there.
  • If you prefer the audio never to leave the NappAI server, use a local STT such as Parakeet (local, multilingual). The differences between local and cloud engines are explained in Quality, noise and latency.