Setup Wizard

🌐 Language: English | Français

Setup Wizard

The setup wizard is the guided first-run flow that walks you through every decision dictee needs before it can transcribe. It covers eight topics in order — ASR backend, shortcuts, translation, optional Docker / Ollama installs, microphone, visualization, and a final test — and writes your answers to ~/.config/dictee/dictee.conf.

On a fresh install the wizard opens automatically the first time you launch dictee-setup. You can re-enter it anytime with dictee-setup --wizard, or open the regular settings window (same binary, no flag) to tweak a single option without replaying the full wizard.

Table of Contents

• When does the wizard run?
• Step-by-step walkthrough
  • 1. Choose an ASR backend
  • 2. Configure the ASR backend
  • 3. Keyboard shortcuts
  • 4. Choose a translation backend
  • 5. Configure the translation backend
  • 6. Install Docker image / Ollama model (if needed)
  • 7. Microphone, visualization, and options
  • 8. Tests
• Re-running the wizard later
• Where answers are stored

---

When does the wizard run?

Situation	Behaviour
First launch after install	Wizard opens automatically
~/.config/dictee/dictee.conf missing	Wizard opens automatically
DICTEE_SETUP_DONE=true already set	Regular settings window, no wizard
dictee-setup --wizard	Forces wizard mode, even after first setup

You can quit at any point with Quit setup — the bits you already answered are saved, but DICTEE_SETUP_DONE stays false so the wizard reopens next time until completed.

Step-by-step walkthrough

The captions overlaid on the GIF above correspond 1-to-1 with the sections below.

1. Choose an ASR backend

Pick the speech recognition engine that will turn your voice into text. dictee ships with five backends: Parakeet-TDT (recommended default), Canary, Vosk, faster-whisper, and Nemotron. Each has a different trade-off on language coverage, latency, accuracy, and VRAM usage.

👉 Full comparison: ASR-Backends · GPU picks: GPU-Setup 👉 Direct per-backend config: Parakeet-TDT · faster-whisper · Vosk · Canary · Nemotron.

2. Configure the ASR backend

Once a backend is selected, the wizard shows its specific options — model choice (tiny/base/small/medium/large for Whisper, v3/v2/multilingual for Parakeet, etc.), language, hotwords, and backend-specific toggles (VAD, beam size, compute type).

👉 Parakeet deep dive: Parakeet-TDT-Deep-Dive 👉 Direct per-backend config: Parakeet-TDT · faster-whisper · Vosk · Canary · Nemotron.

3. Keyboard shortcuts

Bind the global hotkeys that start / stop dictation, toggle meeting mode, call the post-processing test panel, etc. The wizard pre-fills sensible defaults (F9 / Alt+F9 / Ctrl+F9) and lets you capture new combos by clicking the shortcut row.

👉 Full shortcut catalogue + KDE/GNOME quirks: Keyboard-Shortcuts.

4. Choose a translation backend

Translation is optional and off by default. If you want dictee to paste translated text instead of the original, pick one of five backends: Google, Bing, LibreTranslate (self-hosted Docker), Ollama (self-hosted LLM), or Canary (built-in ASR translation, no extra backend).

👉 Comparison + privacy matrix: Translation 👉 Direct per-backend config: Google · Bing · LibreTranslate · Ollama · Canary built-in.

5. Configure the translation backend

Source / target languages, model selection (for Ollama), LibreTranslate URL, or Canary target language pair. If the backend is not installed or not reachable, the wizard shows a red warning and the CLI command needed to fix it.

👉 Translation setup: Translation · Ollama: Ollama-Setup 👉 Direct per-backend config: Google · Bing · LibreTranslate · Ollama · Canary built-in.

6. Install Docker image / Ollama model (if needed)

If you picked LibreTranslate (Docker) or Ollama, the wizard offers to pull the required resources for you:

• LibreTranslate → docker run one-liner with the language packs you need.
• Ollama → ollama pull <model> for the chosen translation or LLM correction model.

If the engine itself isn't installed yet, the wizard shows the CLI install one-liner (curl -fsSL https://ollama.com/install.sh | sh for Ollama, distro-specific for Docker) so you can install in a terminal and come back to the wizard.

👉 Detailed Ollama setup: Ollama-Setup · Docker issues: Troubleshooting 👉 Direct install walkthroughs: Configuration#ollama (3 Ollama states) · Configuration#libretranslate (Docker image).

7. Microphone, visualization, and options

Last configuration step: pick your input device, choose the visual feedback (tray icon only, plasmoid widget, or both), enable auto-unmute, auto-start, meeting mode, and optional services (daemon autostart, passthrough loopback).

👉 UI interfaces: Tray-Icon · Plasmoid-Widget.

8. Tests

The wizard ends with a short live-test panel: speak a sentence, see what dictee transcribes, verify post-processing rules, and confirm the pipeline end-to-end. If anything looks off, the back button takes you to the relevant section without losing your answers.

Re-running the wizard later

Three ways to return:

• dictee-setup --wizard from any terminal — forces wizard mode regardless of setup state.
• Tray icon → Settings → Setup wizard (button on the main config window).
• Delete ~/.config/dictee/dictee.conf and relaunch dictee-setup — simulates a fresh install.

Where answers are stored

All wizard decisions land in a single flat KEY=value file:

~/.config/dictee/dictee.conf

Translation / ASR / LLM toggles, selected models, keyboard shortcuts, microphone name, visual preferences — everything. The file is safe to edit by hand once you're familiar with the keys; dictee-setup will preserve unknown keys on save.

👉 Advanced config reference: CLI-Reference · post-processing rules: Rules-and-Dictionary.

---

Next steps

• Read the ASR-Backends comparison to pick the right engine for your use case.
• Configure Keyboard-Shortcuts that won't clash with your desktop.
• If you hit issues during the wizard, check Troubleshooting.