voicebox-voice-synthesis

What this skill does

# Voicebox Voice Synthesis Studio

> Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection.

Voicebox is a local-first, open-source voice cloning and TTS studio — a self-hosted alternative to ElevenLabs. It runs entirely on your machine (macOS MLX/Metal, Windows/Linux CUDA, CPU fallback), exposes a REST API on `localhost:17493`, and ships with 5 TTS engines, 23 languages, post-processing effects, and a multi-track Stories editor.

---

## Installation

### Pre-built Binaries (Recommended)

| Platform | Link |
|---|---|
| macOS Apple Silicon | https://voicebox.sh/download/mac-arm |
| macOS Intel | https://voicebox.sh/download/mac-intel |
| Windows | https://voicebox.sh/download/windows |
| Docker | `docker compose up` |

Linux requires building from source: https://voicebox.sh/linux-install

### Build from Source

**Prerequisites:** [Bun](https://bun.sh), [Rust](https://rustup.rs), [Python 3.11+](https://python.org), Tauri prerequisites

```bash
git clone https://github.com/jamiepine/voicebox.git
cd voicebox

# Install just task runner
brew install just        # macOS
cargo install just       # any platform

# Set up Python venv + all dependencies
just setup

# Start backend + desktop app in dev mode
just dev
```

```bash
# List all available commands
just --list
```

---

## Architecture

| Layer | Technology |
|---|---|
| Desktop App | Tauri (Rust) |
| Frontend | React + TypeScript + Tailwind CSS |
| State | Zustand + React Query |
| Backend | FastAPI (Python) on port 17493 |
| TTS Engines | Qwen3-TTS, LuxTTS, Chatterbox, Chatterbox Turbo, TADA |
| Effects | Pedalboard (Spotify) |
| Transcription | Whisper / Whisper Turbo |
| Inference | MLX (Apple Silicon) / PyTorch (CUDA/ROCm/XPU/CPU) |
| Database | SQLite |

The Python FastAPI backend handles all ML inference. The Tauri Rust shell wraps the frontend and manages the backend process lifecycle. The API is accessible directly at `http://localhost:17493` even when using the desktop app.

---

## REST API Reference

Base URL: `http://localhost:17493`  
Interactive docs: `http://localhost:17493/docs`

### Generate Speech

```bash
# Basic generation
curl -X POST http://localhost:17493/generate \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Hello world, this is a voice clone.",
    "profile_id": "abc123",
    "language": "en"
  }'

# With engine selection
curl -X POST http://localhost:17493/generate \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Speak slowly and with gravitas.",
    "profile_id": "abc123",
    "language": "en",
    "engine": "qwen3-tts"
  }'

# With paralinguistic tags (Chatterbox Turbo only)
curl -X POST http://localhost:17493/generate \
  -H "Content-Type: application/json" \
  -d '{
    "text": "That is absolutely hilarious! [laugh] I cannot believe it.",
    "profile_id": "abc123",
    "engine": "chatterbox-turbo",
    "language": "en"
  }'
```

### Voice Profiles

```bash
# List all profiles
curl http://localhost:17493/profiles

# Create a new profile
curl -X POST http://localhost:17493/profiles \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Narrator",
    "language": "en",
    "description": "Deep narrative voice"
  }'

# Upload audio sample to a profile
curl -X POST http://localhost:17493/profiles/{profile_id}/samples \
  -F "file=@/path/to/voice-sample.wav"

# Export a profile
curl http://localhost:17493/profiles/{profile_id}/export \
  --output narrator-profile.zip

# Import a profile
curl -X POST http://localhost:17493/profiles/import \
  -F "[email protected]"
```

### Generation Queue & Status

```bash
# Get generation status (SSE stream)
curl -N http://localhost:17493/generate/{generation_id}/status

# List recent generations
curl http://localhost:17493/generations

# Retry a failed generation
curl -X POST http://localhost:17493/generations/{generation_id}/retry

# Download generated audio
curl http://localhost:17493/generations/{generation_id}/audio \
  --output output.wav
```

### Models

```bash
# List available models and download status
curl http://localhost:17493/models

# Unload a model from GPU memory (without deleting)
curl -X POST http://localhost:17493/models/{model_id}/unload
```

---

## TypeScript/JavaScript Integration

### Basic TTS Client

```typescript
const VOICEBOX_URL = process.env.VOICEBOX_API_URL ?? "http://localhost:17493";

interface GenerateRequest {
  text: string;
  profile_id: string;
  language?: string;
  engine?: "qwen3-tts" | "luxtts" | "chatterbox" | "chatterbox-turbo" | "tada";
}

interface GenerateResponse {
  generation_id: string;
  status: "queued" | "processing" | "complete" | "failed";
  audio_url?: string;
}

async function generateSpeech(req: GenerateRequest): Promise<GenerateResponse> {
  const response = await fetch(`${VOICEBOX_URL}/generate`, {
    method: "POST",
    headers: { "Content-Type": "application/json" },
    body: JSON.stringify(req),
  });

  if (!response.ok) {
    throw new Error(`Voicebox API error: ${response.status} ${await response.text()}`);
  }

  return response.json();
}

// Usage
const result = await generateSpeech({
  text: "Welcome to our application.",
  profile_id: "abc123",
  language: "en",
  engine: "qwen3-tts",
});

console.log("Generation ID:", result.generation_id);
```

### Poll for Completion

```typescript
async function waitForGeneration(
  generationId: string,
  timeoutMs = 60_000
): Promise<string> {
  const start = Date.now();

  while (Date.now() - start < timeoutMs) {
    const res = await fetch(`${VOICEBOX_URL}/generations/${generationId}`);
    const data = await res.json();

    if (data.status === "complete") {
      return `${VOICEBOX_URL}/generations/${generationId}/audio`;
    }
    if (data.status === "failed") {
      throw new Error(`Generation failed: ${data.error}`);
    }

    await new Promise((r) => setTimeout(r, 1000));
  }

  throw new Error("Generation timed out");
}
```

### Stream Status with SSE

```typescript
function streamGenerationStatus(
  generationId: string,
  onStatus: (status: string) => void
): () => void {
  const eventSource = new EventSource(
    `${VOICEBOX_URL}/generate/${generationId}/status`
  );

  eventSource.onmessage = (event) => {
    const data = JSON.parse(event.data);
    onStatus(data.status);

    if (data.status === "complete" || data.status === "failed") {
      eventSource.close();
    }
  };

  eventSource.onerror = () => eventSource.close();

  // Return cleanup function
  return () => eventSource.close();
}

// Usage
const cleanup = streamGenerationStatus("gen_abc123", (status) => {
  console.log("Status update:", status);
});
```

### Download Audio as Blob

```typescript
async function downloadAudio(generationId: string): Promise<Blob> {
  const response = await fetch(
    `${VOICEBOX_URL}/generations/${generationId}/audio`
  );

  if (!response.ok) {
    throw new Error(`Failed to download audio: ${response.status}`);
  }

  return response.blob();
}

// Play in browser
async function playGeneratedAudio(generationId: string): Promise<void> {
  const blob = await downloadAudio(generationId);
  const url = URL.createObjectURL(blob);
  const audio = new Audio(url);
  audio.play();
  audio.onended = () => URL.revokeObjectURL(url);
}
```

---

## Python Integration

```python
import httpx
import asyncio

VOICEBOX_URL = "http://localhost:17493"

async def generate_speech(
    text: str,
    profile_id: str,
    language: str = "en",
    engine: str = "qwen3-tts"
) -> bytes:
    async with httpx.AsyncClient(timeout=120.0) as client:
        # Submit generation
        resp = await client.post(
            f"{VOICEBOX_URL}/generate",
            json={
                "text": text,
                "profile_id": profile_id,
                "language": language,
                "engine": engine,
            }
        )
        resp.raise_for_status()
        generation_id = resp.json()["generation_id"]

        # Poll until complete
        for _ in range(120):