Voicebox

Voicebox is an open-source voice studio for cloning voices and turning text (or speech) into audio. It runs on macOS, Windows, and Linux, and it can keep your voice data on your machine. If you want a local-first alternative to hosted voice services, it gives you both a desktop workflow and an API you can connect to.

Key things it does well:

• Voice cloning from short samples. Clone a voice from as little as a few seconds of audio, then reuse that voice for future generations.
• Multiple TTS engines. Generate speech using several built-in engines, each with different strengths (such as expressive control or long-form behavior).
• Local dictation with Whisper. Convert speech to text inside Voicebox, with multiple Whisper model sizes to match your hardware.
• Edit and produce multi-voice stories. Build narratives with a timeline-style editor, trim clips, and mix conversations between characters.
• Audio effects pipeline. Add effects like pitch shift, reverb, delay, and compression, with live preview and saved presets.
• Local or remote inference. Run GPU inference locally on your setup, or connect to a remote machine.
• Local API and agent integration. Expose engines as localhost endpoints and provide an MCP tool so MCP-aware agents can speak in your cloned voices.