Voicebox

Voicebox is an open-source voice studio you install on your computer. It’s for people who want to clone voices and produce speech without relying on a paid cloud API, plus for developers who want to wire speech into apps and agents.

It’s built around local-first workflows, so your voice samples can stay on your machine while you generate audio or transcribe speech. You can also run inference on a remote machine when you need more compute.

Key capabilities you’ll likely care about:

• Near-perfect voice cloning: Create a voice profile from a short audio sample (as little as a few seconds) and reuse it for consistent delivery.
• Multiple TTS engines: Generate speech across several engines, including options aimed at higher quality, faster iteration, or better long-form output.
• Dictation with transcription: Use Whisper locally for speech-to-text, with multiple model sizes for different speed and accuracy needs.
• Refinement and “in-character” rewriting: Clean up transcripts locally and restate text in a cloned voice profile, or generate a fresh line in that persona.
• Stories editor: Build multi-voice narratives by arranging, trimming, and mixing clips along a timeline.
• Audio effects pipeline: Apply pitch shift, reverb, delay, compression, and save effect presets per voice profile.