ElevenLabs Plugin for ElizaOS

High-quality text-to-speech (TTS) and speech-to-text (STT) plugin for ElizaOS using the ElevenLabs API.

Features

• Text-to-Speech (TTS): High-quality voice synthesis with multiple voice models
• Speech-to-Text (STT): Accurate transcription with Scribe v1 model
• Speaker Diarization: Identify up to 32 different speakers
• Multi-language Support: 99 languages for STT
• Audio Event Detection: Detect laughter, applause, and other audio events
• Streaming Support: Efficient memory usage with streaming audio output
• Multi-runtime: Available for TypeScript, Python, and Rust

Installation

TypeScript (npm)

npm install @elizaos/plugin-elevenlabs
# or
bun add @elizaos/plugin-elevenlabs

Python (PyPI)

pip install eliza-plugin-elevenlabs

Rust (crates.io)

Add to your Cargo.toml:

[dependencies]
eliza-plugin-elevenlabs = "0.1.0"

Quick Start

TypeScript

import { elevenLabsPlugin } from '@elizaos/plugin-elevenlabs';

// Add to your character configuration
const character = {
  plugins: ['@elizaos/plugin-elevenlabs'],
  settings: {
    ELEVENLABS_API_KEY: 'your-api-key',
  },
};

Python

from eliza_plugin_elevenlabs import ElevenLabsService

async with ElevenLabsService(api_key="your-api-key") as service:
    # Text-to-speech
    audio = await service.text_to_speech_bytes("Hello, world!")
    
    # Speech-to-text
    transcript = await service.speech_to_text(audio_bytes)

Rust

use eliza_plugin_elevenlabs::ElevenLabsService;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let service = ElevenLabsService::new("your-api-key");
    
    // Text-to-speech
    let audio = service.text_to_speech("Hello, world!").await?;
    
    // Speech-to-text
    let transcript = service.speech_to_text(&audio).await?;
    
    Ok(())
}

Configuration

Project Structure

plugin-elevenlabs/
├── package.json          # Root package with multi-language scripts
├── README.md             # This file
├── .gitignore
├── .github/
│   └── workflows/
│       ├── ci.yml        # CI for all languages
│       ├── npm-deploy.yml    # npm publishing
│       ├── pypi-deploy.yml   # PyPI publishing
│       └── crates-deploy.yml # crates.io publishing
├── typescript/           # TypeScript implementation
│   ├── package.json
│   ├── src/
│   └── README.md
├── python/               # Python implementation
│   ├── pyproject.toml
│   ├── src/
│   └── README.md
└── rust/                 # Rust implementation
    ├── Cargo.toml
    ├── src/
    └── README.md

Development

Build All

bun run build

Build Individual Languages

bun run build:ts      # TypeScript
bun run build:python  # Python
bun run build:rust    # Rust

Test All

bun run test

Test Individual Languages

bun run test:ts      # TypeScript
bun run test:python  # Python
bun run test:rust    # Rust

Lint

bun run lint         # All languages
bun run lint:ts      # TypeScript
bun run lint:python  # Python
bun run lint:rust    # Rust

Model Types

TEXT_TO_SPEECH

Converts text into spoken audio. Supports:

• Multiple voice models
• Configurable voice parameters
• Streaming output
• Various audio formats (MP3, PCM, etc.)

TRANSCRIPTION

Converts audio/video into text transcripts. Supports:

• 99 languages with auto-detection
• Speaker diarization (up to 32 speakers)
• Word/character-level timestamps
• Audio event tagging

Supported Models

TTS Models

• eleven_monolingual_v1
• eleven_multilingual_v1
• eleven_multilingual_v2
• eleven_turbo_v2
• eleven_turbo_v2_5

STT Models

• scribe_v1

License

MIT

Links

• ElevenLabs API Documentation
• ElizaOS Documentation
• GitHub Repository