>

speak.transcribe.
keep flow.

Open-source Rust speech-to-text for GNOME/Wayland.
Native IBus integration, global shortcut dictation, and local model choices.

install with dnfview source
Dikt general settings page

Experimental features can enable live preview during active dictation sessions.

// features

Built for GNOME dictation

Configurable Shortcut Toggle

Press once to start dictation. Press again to stop and commit text.

Local Model Inference

Whisper, Parakeet, and SenseVoice models run locally for speech-to-text.

Experimental Live Preview

Optional beta mode shows a live preedit preview while recording.

Optional AI Cleanup

Coming soon

Post-process transcripts with providers like OpenAI, Anthropic, Groq, and more.

// quickstart

Three steps. Dictating.

01

Install

Install ibus-dikt from the RPM repository.

02

Configure

Pick your model and set your dictation shortcut.

03

Dictate

Press shortcut to start, press again to commit text.

// models

Choose your engine

Real Dikt model manager UI from the current build.

Dikt model management page
# rpm / fedora gnome wayland

Install and start dictating

Dikt currently ships as an RPM package for GNOME/Wayland systems.

DEB support is in progress, with broader GNOME/Wayland distro support planned.

Step 1: Add repository
$ sudo dnf config-manager addrepo --from-repofile=https://rohithmahesh3.github.io/dikt-rpm/dikt.repo
Step 2: Refresh metadata
$ sudo dnf makecache
Step 3: Install package
$ sudo dnf install ibus-dikt
view source

Updates via sudo dnf upgrade