Add selectable Voxtype and Superwhisper dictation (#14552)

* Add selectable dictation backends

* Move dictation migration after repository priority update

* Extract backend-specific dictation installers

* Keep dictation installer dispatcher executable

* Call dictation backend installers directly

* Drop legacy Superwhisper migration and runtime cleanup

* Separate dictation defaults from generic recording protocol

* Add backend-specific dictation removers

* Preserve existing Voxtype selection during upgrade

* Make dictation backend selection safe and retryable

* Show the dictation indicator only for selected Voxtype

* Allow switching away from stopped Superwhisper

* Remove the built-in bar dictation indicator

* Store default dictation selection under defaults

* Rename dictation backend selector to use

* Move dictation removers into their own menu category

manual/07-hotkeys.md

@@ -150,8 +150,8 @@ Usually on Linux, you need `Ctrl + Shift + C/V` to copy'n'paste in the terminal
 | `Super + Alt + ]` | Make webcam overlay larger while recording |
 | `Alt + Shift + L` | Copy current URL from webapp or Chromium |
 | `Alt + Shift + D` | Download the video on the current page to `~/Videos` |
-| `Super + Ctrl + X` | Start/stop dictation (requires _Install > AI > Dictation_) |
-| `F9` | Push-to-talk dictation (requires _Install > AI > Dictation_) |
+| `Super + Ctrl + X` | Start/stop dictation (requires _Setup > Defaults > Dictation_) |
+| `Right Alt` / `F9` | Push-to-talk dictation (requires Voxtype or Superwhisper) |
 
 With screenrecordings, the hotkey first asks which audio you want, then starts recording. Hit it again to stop. See [screenshots and recording](12-screenshots-recording.md) for the details.
 

manual/11-text-extraction-dictation.md

@@ -10,6 +10,14 @@ This is very helpful for grabbing addresses out of image footers or phone number
 
 ### Dictation
 
-Omarchy offers AI dictation via [Voxtype](https://voxtype.io/). You install it via _Install > AI > Dictation_ through the Omarchy menu. By default, it'll load a base English model that takes up 150MB. But you can tweak which model you'd like to use by running `voxtype setup model` in the terminal. And you can tweak all the settings via `~/.config/voxtype/config.toml`.
+Install [Voxtype](https://voxtype.io/) or [Superwhisper](https://superwhisper.com/) through _Setup > Defaults > Dictation_ in the Omarchy menu. Selecting a backend installs it if needed and makes it the default for dictation. Voxtype loads a base English model that takes up 150MB; change the model with `voxtype setup model` and other settings through `~/.config/voxtype/config.toml`. Superwhisper opens its settings to choose cloud or local processing, and OPR keeps the application updated.
 
-Once installed, you dictate by holding down `F9` or by toggling with `Super + Ctrl + X`, and the dictated text will appear in the focused input area.
+Once installed, you dictate by holding down `Right Alt` or `F9`, or by toggling with `Super + Ctrl + X`, and the dictated text will appear in the focused input area. These shortcuts are enabled only after selecting a backend. Right Alt also starts dictation when used in a modifier chord; use Left Alt for those chords. On keyboard layouts where Right Alt is AltGr, use `F9` to keep AltGr available for typing.
+
+The same shortcuts can use an installed Superwhisper instead. Select your backend through Setup > Defaults > Dictation. The backend installer configures it and saves the selection. Run `omarchy default dictation` to see which one is selected. The selection is saved in `~/.config/omarchy/defaults/dictation`. Dictation requires an explicit selection.
+
+Omarchy loads the selected backend's desktop integration automatically. No Superwhisper configuration is needed in `~/.config/hypr/`.
+
+Custom bindings and scripts can use `omarchy dictation start`, `omarchy dictation stop`, and `omarchy dictation toggle`. The selected backend handles the recording and transcription. Selecting Superwhisper disables its native hold shortcut and sets its native toggle to `Alt + Space`, keeping it separate from Omarchy's recording keys. Superwhisper requires at least one native recording shortcut, and its cancellation shortcut stays available.
+
+Additional backends can integrate without changing Omarchy. Provide an executable named `omarchy-dictation-<backend>` on PATH that accepts `start`, `stop`, and `toggle` and returns promptly, with a nonzero status on failure. Its installer configures the provider and calls `omarchy-dictation-use <backend>` after successful setup. This stops the previous backend and saves the selection atomically; `stop` must succeed when the provider is idle or unavailable. Backend names use lowercase letters, digits, and hyphens. Provide `omarchy-install-dictation-<backend>` for setup; `omarchy default dictation <backend>` launches it. Add a Defaults menu entry through an Omarchy menu extension to call that selector. Optional desktop integration can live in `${XDG_CONFIG_HOME:-$HOME/.config}/<backend>/shortcuts.lua`.