Sonora
What it is
Sonora is a private, fully offline speech studio for Windows. It runs AI text-to-speech, speech-to-text transcription and global voice dictation locally, so scripts and recordings never leave your PC.
I built Sonora, a Windows desktop app that brings text-to-speech, speech-to-text and global dictation together in one offline speech studio. Scripts, recordings, transcriptions and voice samples stay on your PC, because the speech engine and models run locally.
Result
Sonora gives Windows 10 and 11 users private voice creation, transcription and dictation in one app. Internet is only needed for setup, model downloads and update checks.
The problem
Most voice tools send your text and audio to a hosted service, need a constant connection, and split TTS, transcription and dictation across separate apps. I wanted one private studio that does all three on your own machine.
What it does
- Text-to-speech: turn scripts into natural speech with local models, a built-in voice library and compatible custom voice samples
- Speech-to-text: transcribe recordings without uploading them to a third party
- Global dictation: press a configurable hotkey and type with your voice into any focused Windows app, with live streaming
- Integrated player to preview and manage generated audio
- Guided first run that sets up a private Python runtime and downloads the default speech model, so no separate Python install is needed
- Signed automatic updates delivered from GitHub Releases
How it's built
- A Tauri desktop app with a React studio UI and a local Python speech backend that runs multiple TTS and STT engines
- Works on CPU, with an NVIDIA GPU recommended for faster processing
- A marketing site on Next.js, TypeScript, Tailwind CSS, shadcn/ui, Framer Motion, GSAP and React Three Fiber, with a download route that resolves the latest installer from GitHub Releases
