Questions, answered.

No. Audio is transcribed locally by Whisper on your own CPU/GPU and then discarded. There’s no server. The only network request the app can make is an optional update check that sends just the app version and your OS — no audio, no transcripts, no identifiers. You can disable it.

It uses OpenAI’s Whisper models. On Apple Silicon it runs large-v3-turbo on the GPU (Apple MLX), typically 10–20× faster than real time. Intel Macs and Windows use faster-whisper on the CPU with a snappier default model; you can switch to a larger model for more accuracy in settings.

fn (the globe key) on macOS, and Left Ctrl on Windows. macOS exposes fn as a real key; Windows doesn’t (it’s handled in keyboard firmware), so we default to Left Ctrl there. You can remap it in settings.

Whisper handles 40+ languages and auto-detects by default. You can also pin a single language to skip detection and shave a little latency.

The builds aren’t code-signed yet (signing certificates cost money for a free project). On macOS, right-click the app → Open. On Windows, choose More info → Run anyway. Signing is on the roadmap. You can verify the SHA-256 checksums published with each release.

Yes — it pastes at the system cursor, so it works anywhere you can type: editors, browsers, chat, email, terminals. Output is Unicode-safe (accents, emoji, CJK).

A local SQLite history of your transcriptions (text, timestamps, word counts) powers the in-app dashboard. It lives in your OS app-data folder and never leaves the device. You can disable history or clear it anytime from the dashboard.

Yes — free and open source, no subscription, no account. The code is on GitHub.

macOS: quit from the menu bar and move the app to the Trash. Windows: use Add/Remove Programs. To remove the model and history too, delete the Voca folder in your app-data directory.