Offline Whisper: on-device macOS dictation

WhisperShortcut can transcribe with Whisper entirely on your Mac. That means no Google API key is required and your audio is processed locally, without sending it to a cloud service for that mode. Cloud Speech-to-Text with Gemini, GPT, Grok, or any audio-capable model on OpenRouter, plus self-hosted transcription endpoints, are still available when configured; see the Privacy Policy.

When to use it

Choose offline Whisper when you want a strong default for dictation, drafts, and notes with minimal network dependence and a clear on-device path for your audio.

Offline Mode: the whole app on-device

One switch in Settings → Privacy & Permissions makes WhisperShortcut device-local. Dictation runs on Whisper, voice-driven editing and chat run on an on-device model that downloads itself (no server needed), Read Aloud uses macOS voices, and requests the app builds to cloud services are blocked at the network layer. Nothing is written to the usage log. It is built for regulated dictation such as patient findings or case notes, where the recording has to stay on the Mac whatever else is configured. Long recordings start transcribing on-device while you still speak.

Offline Whisper and cloud Gemini

You can use cloud transcription and AI features with your own provider API keys, or pair offline Whisper for transcription with in-app features that fit your workflow. A full feature overview is on the Features page; downloads are on Download.