Speechka is a fast, private voice-to-text app for Mac. Hold a hotkey, speak your thought, release — your words paste straight into any app. On-device by default. No subscriptions — buy once, own it.
No clicking, no context switching, no waiting for a server. Your Mac does everything.
Press and hold your chosen combination — default is ⌘⇧; (fully remappable). The HUD pill appears instantly.
Talk naturally. The live waveform dances as your voice streams straight into the local AI model.
Let go. The transcription types itself (or pastes) into the app you were already using. Done.
Audio, transcripts, history, and settings stay local. Transcription runs on-device unless you explicitly switch to a cloud model with your own API key. The only thing that has to touch the internet is a tiny license check — and even that stops after 60 days for lifetime buyers.
Transcription runs on-device using open-source speech models. With the default setup, no audio is uploaded, streamed, or stored on any server.
Prefer a hosted model? Bring your own OpenAI or Groq API key. Audio goes directly from your Mac to that provider — never through our servers.
Record on a plane, in a café with broken Wi-Fi, or during a focus sprint with internet disabled. The app does not block you.
No analytics, no crash reporting, no content tracking. During the free trial, the app reports only the minutes used — never what you said.
A license key unlocks the app. The free trial starts with a one-time email link — no password, no profile, no login screens.
Every capture is saved locally in a searchable daily journal. Grouped by app, timestamped, and yours forever.
No subscriptions. No recurring bills. Buy once, own forever, and the app goes fully offline after day 60.