# Deepgram (/en/ai/platform/models/asr/deepgram)

> For AI agents: see the complete documentation index at [llms.txt](/llms.txt).

> Documentation: Conversational AI.
> Documentation version: Preview.
> This document describes only this version. [Version index](/llms/doc-versions-convoai-guides-v2.txt).


Deepgram provides fast, accurate automatic speech recognition with advanced AI models optimized for real-time streaming and conversational applications across multiple languages.

## Sample configuration

The following example shows how to configure Deepgram as the agent's ASR when you create or update an agent. For a complete example, see [Create an agent and start a session](/en/ai/platform/get-started/start-stop-agent#create-an-agent).

Deepgram supports managed mode and BYOK. In managed mode, Agora provides access to Deepgram, and you don't need a Deepgram account. For BYOK, store your Deepgram API key in a secret set, and refer to it from the agent. For details, see [Use managed mode](/en/ai/platform/build/managed-mode) and [Use your own provider keys](/en/ai/platform/build/use-your-own-keys).

**Python SDK**

```python
from agora_agent import Agent, DeepgramSTT, secret_ref

# client is your configured Agora client
agent = (
    Agent(client=client)
    .with_stt(
        DeepgramSTT(
            api_key=secret_ref("support_prod", "deepgram"),  # omit to use managed mode
            model="nova-3",
            language="en-US",
            keyterm="agora%20convo",
        )
    )
    .with_llm(...)  # configure your LLM vendor
    .with_tts(...)  # configure your TTS vendor
)
```

**TypeScript SDK**

```ts
import { Agent, DeepgramSTT, secretRef } from 'agora-agents';

// client is your configured Agora client
const agent = new Agent({ client })
  .withStt(
    new DeepgramSTT({
      apiKey: secretRef('support_prod', 'deepgram'), // omit to use managed mode
      model: 'nova-3',
      language: 'en-US',
      keyterm: 'agora%20convo',
    }),
  )
  .withLlm(/* configure your LLM vendor */)
  .withTts(/* configure your TTS vendor */);
```

**Go SDK**

```go
// client is your configured Agora client
agent := agentkit.NewAgent(client).
	WithStt(vendors.NewDeepgramSTT(vendors.DeepgramSTTOptions{
		APIKey:   agentkit.SecretRef("support_prod", "deepgram"), // omit to use managed mode
		Model:    "nova-3",
		Language: "en-US",
		Keyterm:  "agora%20convo",
	})).
	WithLlm( /* configure your LLM vendor */ ).
	WithTts( /* configure your TTS vendor */ )
```

**REST API**

- **Managed mode**

  Set `credential.mode` to `managed` in the `asr` block. Agora provides access to Deepgram, and you don't need a Deepgram account.

  ```json
  "asr": {
    "credential": { "mode": "managed" },
    "vendor": "deepgram",
    "model": "nova-3",
    "language": "en-US",
    "params": {
      "keyterm": "agora%20convo"
    }
  }
  ```


- **BYOK**

  Set `credential.mode` to `byok` and `credential.fields.api_key` to a reference to your Deepgram API key in a secret set. In the following example, the key is stored as `deepgram` in the `support_prod` set:

  ```json
  "asr": {
    "credential": {
      "mode": "byok",
      "fields": {
        "api_key": "$secrets.support_prod.deepgram"
      }
    },
    "vendor": "deepgram",
    "model": "nova-3",
    "language": "en-US",
    "params": {
      "keyterm": "term1%20term2"
    }
  }
  ```

## Key parameters

The `asr` block takes the following fields. In the SDKs, `DeepgramSTT` sets them from its options, and passes any other Deepgram parameters from its additional parameters option to `params`.

**asr required**

**`vendor` (string) required**

Set to `deepgram`.

**`credential` (object) required**

How the agent authenticates with Deepgram. Set `mode` to `managed`, or to `byok` with `fields.api_key` set to a `$secrets.<set>.<key>` reference to your Deepgram API key. Get your API key from the [Deepgram Console](https://console.deepgram.com/).

**`model` (string) optional**

The speech recognition model to use. Required in managed mode, where it must be `nova-2` or `nova-3`.

**`language` (string) optional**

The language code for speech recognition, for example `en-US`, `es`, or `fr`. See [supported languages](https://developers.deepgram.com/docs/models-languages-overview) for language codes.

**`url` (string) optional**

A custom WebSocket endpoint for Deepgram's streaming API, for example `wss://api.deepgram.com/v1/listen`. Omit it to use the default endpoint.

**`params` (object) optional**

**`keyterm` (string) optional**

Boost specialized terms and brands. Only compatible with the `nova-3` model.

> **Caution**
>
> The parameters listed on this page are validated for use with the Conversational AI Platform. Required parameters must be provided as documented. Any additional parameters in `params` are passed through directly to Deepgram without validation. For a full list of supported options, refer to the [Deepgram API documentation](https://developers.deepgram.com/docs/getting-started-with-live-streaming-audio).
>
> The following parameters are not passed through and instead use default values set by Agora:
> `callback`, `callback_method`, `channels`, `encoding`, `multichannel`, `sample_rate`