> ## 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.

# Hume AI

> Use the Hume AI text-to-speech plugin in a Zero Runtime pipeline.

Hume AI is a **text-to-speech** plugin. It turns the LLM's reply into expressive,
emotion-aware audio the caller hears.

## Setup

Set your Hume AI API key in the worker environment. Generate a key from the [Hume AI dashboard](https://platform.hume.ai/settings/keys):

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

## Usage

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

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

tts = HumeAITTS()

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

## Parameters

| Parameter     | Type    | Default              | Description                                                                                                                                                                   |
| ------------- | ------- | -------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `api_key`     | `str`   | `None`               | Hume AI API key. Falls back to the `HUMEAI_API_KEY` environment variable when unset.                                                                                          |
| `voice`       | `str`   | `"Serene Assistant"` | Voice name or UUID. Can be a name from Hume's built-in voice library (e.g. `"Serene Assistant"`, `"Male English Actor"`), the name of a custom saved voice, or a UUID string. |
| `speed`       | `float` | `1.0`                | Speaking speed multiplier on a non-linear scale from `0.5` (much slower) to `2.0` (much faster), where `1.0` is normal pace.                                                  |
| `sample_rate` | `int`   | `24000`              | Advertised output sample rate in Hz. Hume AI's API always returns audio at 48 kHz regardless of this value.                                                                   |

Spoken replies support interruptions. When the caller talks over the agent, the runtime
stops the audio and listens again. See the [plugins overview](/plugins/overview) for how
TTS fits into the rest of the pipeline.