The short version
Rekody turns your speech into text on the device in front of you. It has no sign-up, no analytics inside the app, and it never checks in with a license server. We don't run servers that could receive your audio, so your words have nowhere to go but into the app you're typing in.
On your device
Rekody records only while you dictate, transcribes with an on-device speech model, and places the finished text where your cursor is. Your audio never leaves your device.
- The speech model is downloaded once and runs fully offline. No audio or transcript is ever uploaded.
- Your dictation history is a local file on your device. You can delete any entry, or all of them, at any time.
- There are no accounts, no sign-in, and no analytics or tracking. Rekody asks for no personal information.
- If you add API keys for optional cloud engines, they're stored in your system's keychain, never in plaintext on disk.
- The only network calls Rekody makes on its own are update checks against GitHub releases.
Rekody is off in password and other secure fields by design. It won't offer to dictate where the system marks the input as secure.
Rekody's dictation engine is open source (github.com/rekody/rekody, MIT), so you don't have to take our word for any of this.
If you opt into cloud engines
Where Rekody offers optional cloud speech-to-text or LLM engines, you configure them yourself with your own API keys. When you do, your audio or text goes directly from your device to that provider under their terms. There is no Rekody server in the middle, because we don't run any. If you never add a key, nothing is ever sent.
This website
rekody.com is a static site. It uses privacy-friendly, cookie-less analytics (Vercel Analytics) for anonymous page-view counts. No tracking cookies, no ad pixels, no fingerprinting.
Questions
Email hi@rekody.com, or open an issue on GitHub. For security reports, email security@rekody.com or use a private security advisory.