Talk to Your Code.

Yep. Free as in beer, free as in speech, free as in "wait, what's the catch?" There is no catch. MIT licensed. No paywalls. No "premium tier" upsell in 6 months. The whole thing lives on GitHub. Go judge my code.

Absolutely not. Your voice stays on your machine like it's under house arrest. Whisper runs locally, processes locally, and forgets locally. No servers. No logs. Not even a little peek. Pinky promise.

If your computer was built this decade, you're probably fine. Minimum: 4GB RAM and a pulse. Recommended: 8GB RAM + NVIDIA GPU for that buttery-smooth transcription. No GPU? Still works — just thinks a little harder.

Totally. No GPU? No drama. Synthia detects this and switches to CPU mode automatically. The tiny Whisper model runs great on CPU — like a Honda Civic. Reliable. Gets the job done. Won't win races but never lets you down.

Right now, Synthia and Claude Code are basically best friends. But we're not exclusive — Gemini, Codex, and Copilot support is on the roadmap. PRs welcome if you want to play matchmaker.

Open config.yaml and swap in a different Piper voice. Amy's friendly. Ryan's chill. Lessac sounds like your cool coworker. Pick your fighter. Full list in the README.

Not yet — currently Linux only. But macOS and Windows support is coming in v0.3. Yes, we know. Yes, we're working on it. Follow the roadmap so you can say "finally" when it drops.