voicebox

Project Overview

Voicebox is a local-first open-source AI voice studio and a free alternative to ElevenLabs and WisprFlow. It clones voices from a few seconds of audio, generates speech, and enables dictation into any app with a global hotkey.

Key Features

• Voice cloning: zero-shot cloning from a reference sample
• 7 TTS engines: Qwen3-TTS, LuxTTS, Chatterbox, Kokoro, and more
• Supports 23 languages
• Global dictation hotkey: voice input into any text field
• Stories editor: multi-track timeline for conversations and podcasts
• Voice personalities: attach free-form personas to voice profiles
• REST API and built-in MCP server

Target Audience

• Content creators and podcasters
• Developers needing voice cloning and TTS capabilities
• Privacy-conscious users who want local AI voice tools

Quick Start

1. Download the installer for your platform from voicebox.sh
2. macOS (Apple Silicon) download DMG, Windows download MSI
3. Or use Docker: docker compose up
4. Launch and import or record voice samples for cloning
5. Use the global hotkey to start voice input

Core Advantages

• Complete privacy: models and voice data never leave your machine
• Native performance: built with Tauri (Rust), not Electron
• Cross-platform: macOS (MLX/Metal), Windows (CUDA), Linux, AMD ROCm
• Open-source and free with features comparable to commercial products

System Requirements

• OS: macOS (Apple Silicon/Intel), Windows, Linux
• Architectures: x86_64, arm64
• Additional Dependencies: None (Docker optional)

Notes

• Voice cloning and TTS require significant computing resources
• Some advanced features may require GPU acceleration