Voicebox
Voicebox is an open-source AI voice studio that allows users to clone voices, generate speech, dictate into any app, and talk to agents in voices they own. It is a local-first application, meaning that all models, voice data, and captures never leave the user's machine, ensuring complete privacy. The application supports multiple TTS engines, including Qwen3-TTS, Qwen CustomVoice, Qwen VoiceDesign, LuxTTS, Chatterbox Multilingual, Chatterbox Turbo, HumeAI TADA, and Kokoro, and allows users to switch between them per-generation. Voicebox also features post-processing effects, such as pitch shift, reverb, delay, chorus, compression, and filters, and supports unlimited generation length with auto-chunking and crossfade. Additionally, the application includes a stories editor for multi-voice timeline editing, global dictation and voice input, and agent voice output. Voicebox is built with Tauri (Rust) and uses a variety of technologies, including React, TypeScript, Tailwind CSS, FastAPI (Python), and Pedalboard (Spotify) for effects. The application is available for macOS, Windows, and Linux, and can be installed using a variety of methods, including downloading a pre-built binary or building from source. Voicebox is licensed under the MIT License and is free to use and distribute.
- Category
- Voice & Audio
- Pricing
- Open Source
Problem solved
Voicebox solves the problem of providing a comprehensive and private AI voice studio that allows users to clone voices, generate speech, and interact with agents in a secure and flexible way. It addresses the need for a local-first application that does not rely on cloud services, ensuring that sensitive voice data and models are kept private. By providing a wide range of features and technologies, Voicebox enables users to create high-quality voice content and interact with agents in a natural and intuitive way.
Use cases
- Cloning voices from a few seconds of audio
- Generating speech in 23 languages across 8 TTS engines
- Dictating into any text field with a global hotkey
- Giving any MCP-aware AI agent a voice of your choosing
- Creating multi-voice timelines for conversations, podcasts, and narratives
Audience
- Developers
- ML Engineers
- Data Scientists
- Content Creators
Pros
- Complete privacy
- Support for multiple TTS engines
- Post-processing effects
- Unlimited generation length
- Global dictation and voice input
Cons
- Steep learning curve
- Requires significant computational resources
- Limited support for certain languages and accents
Official website: https://github.com/jamiepine/voicebox
Canonical LambdaBase page: https://www.lambdabase.com/tools/voicebox