Display server
Wayland does not hand an application the global shortcut, and Cloud does not send the paste for you there.
- Text into the focused field
- Pastes itself — install xdotool
- Global shortcut
- Held by the app
- Extra package needed
- xdotool
A floating bubble on your desktop and a push-to-talk shortcut, with the transcription done on our servers. Installs as a .deb, an .rpm or an AppImage, and runs on hardware that cannot carry a local speech model.
SHA-256226c02e798f6c0ee214f3314d2dab96d3eda8f55c7fe7dea175d1a42a9f9349c.sha256 file
lven-cloud_0.1.26_amd64.deblven-cloud_0.1.26_amd64.debAn active Cloud plan is required for transcription. For the on-device build, see Lven for Linux.
sudo apt install ./lven-cloud_0.1.26_amd64.debDownload lven-cloud_0.1.26_amd64.deb
SHA-256226c02e798f6c0ee214f3314d2dab96d3eda8f55c7fe7dea175d1a42a9f9349c.sha256 file
Most cloud dictation apps do not ship for Linux at all. Lven Cloud does, in the same three package formats as the on-device build, with the same bubble.
Price. From $3.49/mo billed yearly, $7.99 month to month, no word cap on a paid plan.
Punctuated, and not kept: the transcript is delivered as you said it unless you turn on AI polish, and neither audio nor text is kept after delivery.
On X11 the app pastes into the focused field itself (install xdotool). On Wayland the transcript lands on the clipboard and you paste with Ctrl+V, and the toggle is bound in your compositor.
Wayland does not hand an application the global shortcut, and Cloud does not send the paste for you there.
Everything the app needs ships inside the package. What has to come from your distribution is glibc 2.34 or newer — Ubuntu 22.04 and anything later, Debian 12, Fedora 35.
The standard Linux audio stack. Any microphone your audio server already recognizes.
Your voice is sent encrypted and the text comes back, so there is no speech model to download. It needs a connection while you dictate.