Press a key. Speak. It's typed.

WhisperShortcut is a macOS menu bar app for voice-first productivity. Press a shortcut, speak, and your words land on the clipboard (the direct-download build can also paste them straight at the cursor). Longer recordings are transcribed while you still speak, so the text is ready moments after you stop. Use Google Gemini, GPT, Grok, or any audio-capable model on OpenRouter in the cloud, or Whisper fully offline. No account, no subscription.

Yes. WhisperShortcut runs Whisper locally on your Mac, so transcription works without internet and without any API key. One Offline Mode switch makes the whole app device-local: dictation runs on Whisper, voice editing and chat run on an on-device model, Read Aloud uses macOS voices, and requests to cloud services are blocked. Your audio never leaves the device.

Yes. WhisperShortcut is BYOK (bring your own key): you add your own Google, OpenAI, xAI, Anthropic, or OpenRouter API key and pay those providers directly at cost. There is no subscription and no middleman backend.

WhisperShortcut is not locked to one built-in engine. You bring your own API keys and pick whichever model is currently best on the market — Google Gemini, OpenAI GPT, xAI Grok, or Anthropic Claude in chat — and switch the moment a stronger one ships. On top of plain transcription it adds AI on your voice: voice-driven text editing rewrites clipboard text from a spoken instruction, Read Aloud speaks answers back, and there is a built-in AI chat. You can also stay fully offline with Whisper.

You copy some text, press a shortcut, and speak an instruction like "make this more formal." WhisperShortcut sends both to an AI model and puts the rewritten result on your clipboard — voice-driven editing without typing a prompt. Drafts of emails and messages follow your own writing style, learned from examples of messages you actually wrote, and your most frequent instructions can be run with one key instead of speaking. (In the app this is called Dictate Prompt.)

Yes. Highlight any text — or a chat reply — press the Read Aloud shortcut, and WhisperShortcut speaks it back using Gemini, GPT, or Grok voices, or on-device macOS voices offline. Playback starts after the first sentence while the rest streams in, and a small player lets you pause, skip, and change the speed live. You can pick a voice per provider and optionally let AI rewrite the text for more natural spoken output.

Yes. Live Meeting records and transcribes in real time using chunked audio, so long meetings are transcribed continuously rather than in one giant upload. While the meeting runs, short notes are written into the chat as things are said, and you can ask the chat about anything said since the meeting started — with one-tap questions for Catch me up, Action items, Open questions, and Decisions. A shortcut flags an important moment so the final summary is written around it, and transcripts stay in a local folder on your Mac.

Cloud dictation runs at temperature 0.0 by default, which tells the model to reproduce what you said verbatim instead of paraphrasing it. You can raise the temperature, and on Gemini raise the thinking effort, in Settings when audio is hard, accented, or full of unusual vocabulary. A glossary lets you fix names and terms the model keeps getting wrong, including homophone tie-breakers.

Google Gemini, OpenAI GPT, and xAI Grok for cloud transcription, prompting, and chat, plus Anthropic Claude models in chat — and offline Whisper for transcription. Dictation can also run through OpenRouter, so a single OpenRouter key gives you any audio-capable model it routes to, chat can run an on-device model with no server at all or point at a local OpenAI-compatible server such as Ollama or LM Studio, and chat and dictation can use your own Azure OpenAI or Google Vertex AI tenant. You bring your own API keys and switch models anytime.

Your content never passes through a WhisperShortcut server, and there is no telemetry unless you opt in to anonymous usage statistics (counts only, off by default) — requests go directly from your Mac to the provider you chose (Google, OpenAI, xAI, Anthropic, or OpenRouter), or to your own Azure or Vertex tenant, or stay fully local with Offline Mode. No account is required.