use cases
Every one of them starts at the same floating button. Press a shortcut and talk, and the words land wherever your cursor is. Or drag over part of the screen, draw on it while you explain, and the picture and your words arrive together. Pick the one that sounds like your day.
Press Ctrl ` and speak. The text is typed at your cursor, in whatever app has it — a chat, a draft, a document, a terminal.
Speak the prompt, read it over, send it — and when the question is about something on screen, crop that and ask.
for anyone who types a lot of contextOne shortcut covers Claude.ai, Claude Desktop and Claude Code in the terminal. With Pro, Claude can ask you for a snip.
for Claude and Claude Code usersReplies at talking speed in Gmail, Outlook or the mail app on your desktop. Nothing goes out until you press Send.
for whoever owns the inboxFirst drafts into Word, Google Docs, Notion or a plain text file — no word caps, and it works on the train.
for writers, note-takers, anyone facing a blank pageExplain the topic out loud and it becomes your notes; box a diagram and ask AI about it. Free, no student plan needed.
for studentsMove the bulk of the day's typing to your voice — unlimited, because when it is your main input a cap is a wall.
for sore wrists and RSIPress Alt ` , drag over the part that matters, and mark it up while you talk. The annotated screenshot and your transcribed words land together wherever you paste.
Crop the error, circle it while you ask, paste — the screenshot and your question arrive in ChatGPT or Claude together.
for any question about something you can seeWhat broke is in the picture, how it happened is in your words, and both land in the tracker in one paste.
for QA, developers and everyone who finds bugsDraw on the draft while you say what should change — designs, documents, slides, PDFs, live pages.
for reviewers, editors and clientsAnswer tickets by voice and send the customer a screenshot with the button already circled.
for support teamsDictation and core Crop & Speak are free: no word caps, no account, no trial clock, and free for commercial use. A one-time €39 Pro upgrade adds power features on top, and nothing that is free today ever moves behind the paywall.Free vs Pro →
The speech model downloads once, then everything runs on your computer. Pull the network cable and dictation keeps working.
Speech becomes text on your device and screenshots stay in a local folder. Audio is never uploaded, the app tracks nothing about you, and cloud AI exists only as an opt-in through your own accounts.The full network-use table →
Polish fixes grammar, Prompt turns a ramble into a clear request, and you can add your own. They run on the built-in local model or your own Ollama server; Pro can route them through your own OpenAI or Claude account.
Ctrl+` starts dictation, Alt+` starts Crop & Speak, Alt+1 captures without voice. Change any of them in Settings.
One winget command or a signed installer. The app is built cross-platform; the macOS and Linux builds are being validated and will appear on the download page the moment they are ready.Download →
Want the whole feature list first? Every feature in detail · speech to text for Windows · fair questions answered.