> ## Documentation Index
> Fetch the complete documentation index at: https://docs.praison.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Deepgram

> Real-time transcription with Deepgram

Professional speech-to-text with Deepgram's Nova models.

## Setup

```bash theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
export DEEPGRAM_API_KEY=your-key
```

## Usage

```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import AudioAgent

agent = AudioAgent(llm="deepgram/nova-2")
text = agent.transcribe("audio.mp3")
print(text)
```

## Models

| Model                    | Description           |
| ------------------------ | --------------------- |
| `deepgram/nova-2`        | Latest, best accuracy |
| `deepgram/nova`          | Previous generation   |
| `deepgram/whisper-large` | Whisper via Deepgram  |

## With Language

```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
text = agent.transcribe("audio.mp3", language="en-US")
```

<Tip>
  Deepgram excels at real-time streaming and domain-specific recognition.
</Tip>