Skip to content

Setup Wizard

Chris Smashe edited this page Jun 23, 2026 · 5 revisions

Setup Wizard

The Setup Wizard is a guided, six-step onboarding flow that takes you from a fresh install to a working dictation setup. It runs automatically the first time you launch TypeWhisper, and you can reopen it any time from the Open Wizard button in the top-right of every page.

The Setup Wizard, step 1 of 6 — choosing a transcription model (the recommended local whisper.cpp model) with an optional writing-focus vocabulary pack, and the six-step progress dots

What it does

The wizard walks through six steps:

  1. Pick a transcription model — choose your speech-to-text model, with a recommended default. If the model still needs downloading, the wizard downloads and loads it before you continue. This step also has a Writing focus picker — General, Real Estate, Architecture, or Legal — that enables a matching industry vocabulary term pack when you finish. See Transcription engines.
  2. Review extension plugins — see the available plugins and their enable state, so you can turn on the transcription engines, LLM providers, and other extensions you want. See Plugins.
  3. Confirm hotkey and microphone — set your main dictation hotkey and pick your input device. See Global hotkeys.
  4. Setup checklist — a machine-driven checklist of setup tasks tailored to your desktop and session (see below).
  5. First-dictation check — a quick try-it step that confirms dictation works end to end. It includes a microphone test with a live input level meter, a paste smoke test (it types the known string typewhisper paste test into a field and verifies it landed), and a record-and-transcribe check.
  6. Done — finishes onboarding.

Wizard step 2 of 6 — Extensions: enable the transcription engines, LLM providers, and other plugins you want (cloud ones need an API key added later from the Plugins tab)

The setup checklist

Step 4 is the heart of the wizard on Linux. It runs a set of setup tasks that each self-gate on your detected desktop and session, so the same wizard configures GNOME/Wayland, KDE, Hyprland, Sway, and X11 without you having to pick your environment. The tasks cover:

  • Clipboard tooling — the clipboard backend for your session (xclip on X11, wl-copy / wl-paste on Wayland). See Text insertion.
  • Automatic paste — the typing/paste backend (wtype, ydotool, or xdotool) and its setup, including the ydotool daemon and input-group membership where needed. See Text insertion and Wayland notes.
  • Global-hotkey registration — registering your dictation hotkey so it fires from any window. See Global hotkeys.
  • Active-window detection — the compositor-native provider (or the GNOME Window Calls extension) used for profile matching. See Profiles and Wayland notes.
  • ffmpeg — for importing audio/video formats on the File transcription page.
  • CUDA check (optional) — only shown when an NVIDIA GPU is detected; runs a quick check to confirm GPU acceleration is usable. Machines without an NVIDIA GPU have nothing to check, so the button stays disabled.

Wizard step 4 of 6 — the setup checklist: clipboard helper, automatic paste, global dictation shortcut, active-window detection, and audio conversion (ffmpeg) all verified, plus an optional CUDA check and a Re-check button

Required tasks gate the Finish button — you cannot finish until they report ready, though you can Skip to bypass the gate and finish anyway. Tasks that do not apply to your session are skipped automatically.

The global-hotkey task installs the keyboard uaccess udev rule, which grants your active session keyboard access immediately — no logout needed. On init systems without systemd-logind it falls back to adding you to the input group instead; in that case the wizard shows a re-login notice, because group membership only takes effect after you log out and back in. See Wayland notes.

Reopening the wizard

The wizard appears on first launch and marks onboarding complete once you finish. To run it again later — after switching desktops, reinstalling, or installing missing helpers — click Open Wizard in the top-right of any page.

Related pages


Changelog

Date Change
2026-06-17 Initial version.
2026-06-17 Added Setup Wizard screenshot.
2026-06-17 Added wizard plugins (step 2) and setup-checklist (step 4) screenshots.
2026-06-17 Documented the step 1 Writing-focus vocabulary picker, the step 5 mic test and paste smoke test, the optional CUDA check, and the input-group re-login notice.
2026-06-23 v0.10.0: the global-hotkey task now installs the keyboard uaccess rule (immediate, no logout); the re-login notice applies only to the non-logind input-group fallback.

Clone this wiki locally