Use OpenWhisper
Dictation & audio files
Use shortcuts, auto-paste, file uploads, and your saved transcript history.
On this page
Use Quick Record or a shortcut for live dictation, and Upload File for recordings you already have. Both use your selected speech engine.
Record from another app#
Set your microphone in Settings → Dictation → Recording and your shortcut in Settings → App → Hotkeys. Focus the app where you want the text, then press the recording shortcut.
- Toggle: press once to start and again to stop.
- Push and hold: hold the shortcut while speaking and release it to stop. Double-tap it to keep recording hands-free, then press again to stop.
The recording overlay opens next to your pointer. Drag it anywhere if it's in the way; it stays there until the recording finishes, and never takes focus from the app you're dictating into. When you stop, OpenWhisper transcribes and follows your clipboard and auto-paste preferences. Supported engines show a live preview while you speak.
If a transcription fails, the recording isn't lost: History keeps the audio with Retry transcription, and a successful retry fills in the same entry, ready to copy.
Default shortcuts#
Change these in Settings → App → Hotkeys.
| Action | Windows / Linux | macOS |
|---|---|---|
| Start / stop | Numpad * | Control+Option+R |
| Cancel | Numpad - | Control+Option+Escape |
| Enable / disable program | Ctrl+Alt+Numpad * | Control+Option+Shift+R |
| Minimize to tray | Ctrl+Alt+M | Control+Option+M |
| History Calendar | Ctrl+Shift+D | Command+Shift+D |
| Paste latest walkthrough | Ctrl+Shift+F12 | Control+Shift+F12 |
If your keyboard has no number pad, assign a different recording shortcut. Mouse side buttons (Mouse 4 and Mouse 5) can be shortcuts on Windows and X11.
These have optional shortcuts with no default: Command Mode, Scratchpad, language switching, pasting the original of the last dictation, Meeting Mode, Walkthrough and Screenshot, and Route by voice (see Actions & automation). Profile and action shortcuts always toggle, whatever your recording mode.
The Hotkeys page lists every shortcut in one place: OpenWhisper's own, the walkthrough ones, and each profile, transform, and action, the Agent hotkey included. Click a box to set one or Clear to remove it. A row says when its shortcut does nothing because something else uses it or the action is off. Reset to defaults leaves your profile and action shortcuts alone.
Microphones and languages#
Rank microphones in Settings → Dictation → Recording to prefer a device and fall back automatically when it disappears.
Choose Languages I dictate in in Settings → Dictation → Voice model. With two or more, switch with the overlay chip, the tray Language menu, or a shortcut, even mid-dictation. Choices depend on the engine; Parakeet MLX uses Auto and Moonshine is English-only. This selects the recognition language; it doesn't translate.
Configure auto-paste and the clipboard#
Settings → App → General controls output:
- Auto-paste types the finished text into the app you started in. On macOS, Set up auto-paste… names the app that needs Accessibility permission.
- Restore the clipboard after pasting (off by default) puts back whatever you had copied once the transcript lands. Left off, the transcript stays on the clipboard so you can paste it again. If you turned it on before 2.8.3, it stays on.
On X11 Linux, global shortcuts and auto-paste work everywhere. Omarchy supports desktop shortcuts and auto-paste on Hyprland 0.55+. Other native Wayland desktops use focused-window shortcuts and manual paste.
If a shortcut or auto-paste fails, try the in-app button and manual copy first, then follow Troubleshooting.
Transcribe existing files#
Open Upload File and drop in one audio file or a queue. Keep the results separate or combine them. Upload results stay in the app with Copy buttons; they don't auto-paste. Stop keeps files that already finished and marks the rest.
If a file can't be transcribed, the card says why in plain words, for example that the file is empty or isn't audio. Choose an engine that supports uploads in Choosing a model; a cloud engine sends the audio to that provider.
Find and export past transcripts#
History lets you search saved transcripts, play saved audio, retranscribe recordings, and export Markdown, plain text, or JSON. When AI cleanup changed an entry, you can switch between its original and cleaned text. Walkthroughs and actions appear here too, labelled Walkthrough and Action · name.
View → History Calendar (Ctrl+Shift+D) lays dictations, uploads, and meetings out on a month grid. Click the month title or press M to jump to any month or year.
Use plain text for a readable transcript, Markdown when you want formatting, and JSON when you need structured fields. Exports reference recordings but don't include the audio files, so they aren't a recording backup. The export format reference lists every field.
Scratchpad, Stats, and backups#
View → Scratchpad opens a floating notepad. Dictation goes into it while it's focused, and notes save automatically.
Stats in the main window (also View → Stats and the tray menu) shows your words, time saved over typing, pace, cleanup changes, streaks, when you dictate, and top apps for the last 7 days, the last 30 days, or all time, plus your words by kind of app, cleanup level, and language. Across OpenWhisper has a row for each feature you've used (meetings, audio files, walkthroughs, actions, commands, snippets, and dictionary corrections) with a small trend; click a row, or focus it and press Enter, to open that feature. Only dictated words count toward streaks and milestones. Stats update while open, stay local, and survive Clear history and deleting a meeting; Reset stats starts over.
Settings → Models & storage → Backup & restore makes a .owbackup archive with settings, history, meetings, and Scratchpad notes. Include saved recordings and captures (on by default) adds your audio and walkthrough documents. See the backup guide.
Get help from the community, or tell us what went wrong.