Click the button, speak, click again. Your words are typed wherever your cursor is, in 25 languages, with nothing to set up and nothing to switch. Recognition runs on your own CPU — no cloud, no account, no audio sent anywhere.
Free & open source · GPL-3.0 · No telemetry, ever
One button, three moments, about a third of a second of thinking.
Click the floating button, or the icon in your tray, from any app: email, editor, browser, terminal.
Say the thing. A quiet tone marks the start, the button shows a live waveform while it listens, and a turning arc while it thinks.
Your words appear where the cursor is, filler words trimmed and your own spellings applied — and stay on the clipboard, in case they landed nowhere.
Speech recognition runs on your CPU using NVIDIA's open Parakeet model. After the one-time model download, the app pins itself offline — unplug the internet and Talkmorphic will not notice. It makes exactly two kinds of network request in its whole life: fetching the model once, and checking for updates when you ask it to.
Click the small button beside it, type what it misheard and the right spelling once. Never see that mistake again. Your dictionary exports and imports as a file.
The words are left on the clipboard as well as typed. If you clicked away before speaking and the text lands nowhere, right-click and paste — it is still there. A quiet tone marks the start and end of every dictation, so you know the microphone is open without watching the button.
From English, French and German to Polish, Greek and Ukrainian — it hears which one you are speaking and writes it, with nothing to set. One language per dictation, so start a new one to switch. Talkmorphic's own buttons and messages come in the same twenty-five.
Claude Code has voice input built in. Talkmorphic is clearer and more accurate in everyday use, and it is not confined to one window: it types into an email, a terminal, a form in the browser, anywhere your cursor is. Built and used daily by someone who dictates for a living.
A single AppImage. Mark it executable and run it. It updates itself from Settings, only when you press the button.
One download, two very different jobs underneath. Typing into another app is something a Wayland compositor guards closely, and rightly, so there Talkmorphic asks through the standard desktop portal and the compositor stays in charge — no ydotool, no joining the input group, no udev rules. On X11 no permission exists to ask for, so it simply types. Talkmorphic works out which it is on and picks its own way in; you are told plainly if a desktop cannot support it.
Keyboard shortcuts are gone. On Wayland the compositor's press and release signals proved unreliable enough to lose whole dictations, and one habit that works everywhere beats two that differ. So there is a floating button you can drag anywhere, plus the tray icon. A click has nothing to go wrong.
On X11 you can highlight a misheard word and open the correction popup with it already filled in. Wayland does not let a background app read your selection — deliberately — so there the popup asks you to type the word instead.
Talkmorphic runs on Wayland and on X11. It is one AppImage; there is nothing to choose between.
What it needs depends on which you are in. To see which:
echo $XDG_SESSION_TYPEwayland — your desktop must provide the RemoteDesktop portal, which is how Talkmorphic is allowed to type: GNOME 45+ or KDE Plasma 5.27+. You approve it once, on first run.x11 — any desktop at all, including Cinnamon, XFCE, MATE and bare window managers. Nothing to install and nothing to approve.Talkmorphic says so plainly if a desktop cannot support it, rather than failing quietly. You will also need PipeWire or PulseAudio for the microphone, and a 64-bit (x86_64) machine.
No. The speech model runs on your own processor. After the one-time download of the model on first run, Talkmorphic makes no network requests at all except when you ask it to check for an update.
Both, from the same download. On Wayland it types through the desktop's RemoteDesktop portal, which needs GNOME 45 or later, or KDE Plasma 5.27 or later, and which you approve once. On X11 it types directly and works on any desktop, including Cinnamon, XFCE, MATE and bare window managers.
Yes, and there is nothing to set: the speech model settles on whichever one you start speaking. The 25 are Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish and Ukrainian. It handles one language per dictation, so switching part-way through a sentence will not work — stop and start a new one instead. Talkmorphic's own buttons and messages come in the same 25, picked on the first screen and changed at any time in Settings.
Click the floating button, speak, then click it again. There are no keyboard shortcuts: a compositor's key press and release signals proved unreliable enough to lose whole dictations, and a click has nothing to get wrong.
They are on the clipboard: right-click and paste. The commonest cause is that nothing was in focus when you spoke, so there was no field for the text to land in. Talkmorphic leaves every transcript on the clipboard for exactly this reason, and you can switch that off in Settings if you would rather keep whatever you had copied. Anything dictated is also in History, unless you have turned History off.
About 1.4 GB in total. The speech model is roughly 600 MB and is downloaded once on first run; the application itself is a single 97 MB file.
Open Settings and click the small coloured dot beside the version number. Grey while it looks, green if you are current, amber if there is a newer version — click again to download it, and once more when it turns blue to restart into it. Red means GitHub could not be reached; hover for the reason. Nothing is ever downloaded or installed unless you press that dot, and Settings has a Help page explaining the colours.
Settings has a Remove Talkmorphic completely button. It deletes the speech model, your settings, history and dictionary, the menu and autostart entries, and the application file itself. Nothing is left behind.
Mark the file executable (right-click → Properties → “Allow executing” on most desktops), then open it. On first run it downloads the speech model once, about 600 MB, and works fully offline afterwards. Your desktop will ask once for permission to type into other windows — that is the portal doing its job.