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

# NVIDIA

> Use the NVIDIA Riva text-to-speech plugin in a Zero Runtime pipeline. Setup and usage in Python.

NVIDIA is a **text-to-speech** plugin. It converts the agent's text responses into
natural-sounding audio using NVIDIA's GPU-accelerated Riva voices.

## Setup

Set your NVIDIA API key in the worker environment. Generate a key from the [NVIDIA build portal](https://build.nvidia.com/):

```bash theme={null}
export NVIDIA_API_KEY=<key>
```

## Usage

Import the plugin and pass it to the pipeline's `tts` slot.

```python theme={null}
from zrt.plugins import NvidiaTTS

tts = NvidiaTTS()

# Pipeline(tts=tts, ...)
```

## Parameters

| Parameter       | Type          | Default                 | Description                                                                                             |
| --------------- | ------------- | ----------------------- | ------------------------------------------------------------------------------------------------------- |
| `api_key`       | `str \| None` | `None`                  | NVIDIA API key. Falls back to the `NVIDIA_API_KEY` environment variable when unset.                     |
| `voice`         | `str`         | `"English-US-Female-1"` | Riva voice name, e.g. `"English-US-Female-1"` or a Magpie voice like `"Magpie-Multilingual.EN-US.Mia"`. |
| `language_code` | `str`         | `"en-US"`               | BCP-47 language tag for synthesis.                                                                      |
| `sample_rate`   | `int`         | `22050`                 | Output audio sample rate in Hz.                                                                         |
| `server`        | `str`         | `""`                    | Riva gRPC endpoint.                                                                                     |

The synthesized audio is streamed back to the caller as soon as it's generated, after the [LLM](/plugins/llm/google) produces its response.