ElevenLabs Rust SDK

Unofficial Rust SDK for the ElevenLabs API. Provides typed access to all REST endpoints and WebSocket streaming.

Features

• Full API Coverage: 220+ endpoints across Text-to-Speech, Speech-to-Text, Voices, Dubbing, Studio, Conversational AI, and more
• WebSocket Streaming: Real-time TTS input-streaming and Conversational AI via WebSocket
• Type-Safe: Strongly typed request/response types generated from the official OpenAPI spec
• Async/Await: Built on tokio and hpx for async HTTP
• Automatic Retries: Configurable retry with exponential backoff
• Error Handling: Typed errors with status codes and rate-limit info

Installation

Add to your Cargo.toml:

[dependencies]
elevenlabs-sdk = "0.1.0"
tokio = { version = "1", features = ["rt-multi-thread", "macros"] }

Quick Start

use elevenlabs_sdk::{ClientConfig, ElevenLabsClient};
use elevenlabs_sdk::types::TextToSpeechRequest;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Create client from the ELEVENLABS_API_KEY environment variable
    let config = ClientConfig::from_env()?;
    let client = ElevenLabsClient::new(config)?;

    // List voices
    let voices = client.voices().list(None).await?;
    for voice in &voices.voices {
        println!("{}: {}", voice.voice_id, voice.name);
    }

    // Text-to-speech
    let request = TextToSpeechRequest::new("Hello, world!");
    let audio = client
        .text_to_speech()
        .convert("21m00Tcm4TlvDq8ikWAM", &request, None, None)
        .await?;
    std::fs::write("output.mp3", &audio)?;

    Ok(())
}

Available Services

WebSocket Streaming

use elevenlabs_sdk::{ClientConfig, TtsWebSocket, TtsWsConfig};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let config = ClientConfig::from_env()?;

    let ws_config = TtsWsConfig {
        voice_id: "21m00Tcm4TlvDq8ikWAM".into(),
        model_id: "eleven_turbo_v2".into(),
        voice_settings: None,
        generation_config: None,
        output_format: None,
    };

    let mut ws = TtsWebSocket::connect(&config, &ws_config).await?;

    ws.send_text("Hello from real-time streaming!").await?;
    ws.flush().await?;

    while let Some(resp) = ws.recv().await? {
        if let Some(ref audio) = resp.audio {
            println!("Received audio chunk: {} chars base64", audio.len());
        }
        if resp.is_final == Some(true) {
            break;
        }
    }

    ws.close().await?;
    Ok(())
}

Configuration

Environment Variables

Builder Pattern

use elevenlabs_sdk::{ClientConfig, ClientConfigBuilder, ElevenLabsClient};

let config = ClientConfigBuilder::default()
    .api_key("your-api-key")
    .base_url("https://api.elevenlabs.io")
    .timeout(std::time::Duration::from_secs(60))
    .max_retries(5)
    .build()?;

let client = ElevenLabsClient::new(config)?;

Examples

Run the bundled examples with your API key:

ELEVENLABS_API_KEY=your-key cargo run -p elevenlabs-sdk --example text_to_speech
ELEVENLABS_API_KEY=your-key cargo run -p elevenlabs-sdk --example voices
ELEVENLABS_API_KEY=your-key cargo run -p elevenlabs-sdk --example streaming
ELEVENLABS_API_KEY=your-key cargo run -p elevenlabs-sdk --example websocket_tts

Development

just sdk-test             # Run SDK unit tests
just sdk-lint             # Run clippy
just sdk-doc              # Generate docs
just sdk-build-examples   # Build all examples
just sdk-check-coverage   # Check endpoint coverage vs OpenAPI spec
just sdk-test-integration # Run integration tests with Prism mock server

Project Structure

crates/elevenlabs-sdk/
├── src/
│   ├── lib.rs          # Public API re-exports
│   ├── client.rs       # ElevenLabsClient with HTTP + retry logic
│   ├── config.rs       # ClientConfig builder
│   ├── auth.rs         # API key handling
│   ├── error.rs        # Error types
│   ├── middleware.rs    # Request middleware
│   ├── services/       # Typed endpoint wrappers (one module per API group)
│   ├── types/          # Request/response structs from OpenAPI spec
│   └── ws/             # WebSocket streaming (TTS, Conversational AI)
├── examples/           # Runnable examples
└── tests/              # Integration tests

License

Apache-2.0