Voicebox

Voicebox is an open-source, local-first AI voice studio that runs on a user’s machine. It combines text-to-speech, voice cloning, preset voices, dictation, transcription, captures, voice personalities, a stories editor, and integrations for MCP-aware agents. The app works offline without an account; optional Cloud and Studio tiers add encrypted backup and sync rather than being required for local generation.

✨ Key Features

• Local text-to-speech — Generate speech with supported local engines and reusable voice profiles.
• Zero-shot voice cloning — Create a profile from roughly 10–30 seconds of clear reference audio, subject to consent and engine support.
• Preset voices — Use curated voices when cloning is unnecessary or inappropriate.
• Dictation — Hold a push-to-talk shortcut, transcribe speech with Whisper, optionally refine it locally, and paste text into the focused field.
• Captures and transcripts — Keep original audio alongside transcripts and reuse a capture as a voice sample.
• Voice Personalities — Let a profile rewrite text in character before it is spoken.
• Stories editor — Build multi-voice narratives and longer-form spoken content from reusable profiles.
• MCP and REST integration — Let local agents speak, transcribe audio, and list profiles or captures through the built-in interfaces.
• Privacy-first processing — Local data stays on the machine by default; optional cloud objects are encrypted on-device before upload.
• Remote and hardware options — Use documented GPU acceleration or remote mode when local hardware is insufficient.

💳 Pricing

• Local — $0 forever, no account required; includes local voice cloning, dictation, captures, MCP/personality features, and unlimited local generations.
• Cloud — Coming soon; the reviewed pricing page lists a $12/year launch price for encrypted backup/sync, 25 GB storage, up to five devices, and 30-day version history. Pricing and limits are not final.
• Studio — Coming soon; the page lists $48/year as placeholder pricing for 250 GB storage, unlimited devices, one-year history, and priority support. Pricing is explicitly subject to change.

Pricing note — Voicebox states that the paid tiers are optional cloud services and that local functionality remains free and open source. Cloud data is encrypted before upload, but users remain responsible for voice-consent, copyright, and local-device security.

⚖️ Alternatives Compared

🎯 Use Cases

• Private dictation：Turn speech into text without sending audio to a cloud service.
• Voice prototyping：Test narration with a consented cloned or preset voice before recording.
• Agent interfaces：Let Claude Code, Cursor, or another MCP-aware agent speak locally.
• Course and accessibility content：Generate reusable narration and provide an alternate voice interface.
• Stories and games：Create multi-voice character dialogue with personality prompts.
• Podcast and media drafts：Capture, transcribe, revise, and replay spoken ideas.
• Developer experimentation：Integrate local speech and transcription through MCP or REST while controlling the data boundary.