A small release focused on visibility: you can now always see what is happening to your dictation and where it is processed.
- The recording badge no longer disappears too early. After you stop recording (F8), the badge shows the transcription stage — with the provider's name — then the AI-cleanup stage, and vanishes exactly when the text is inserted. No more "dead gap" between releasing the key and the text landing.
- See where your voice is transcribed. The dictation entry in the tray now shows the selected backend:
Local,NSCloud,OpenAIorCustom— just like the AI assistant already does. - "Install models…" right from the tray. A new entry in the Local model submenu opens the console's model catalog in one click.
- Clearer menu labels. "Language" → "App language", "Model" → "Local model".
- Version and plan in the tray header (e.g.
v0.5.2 · Beta). The title greys out while paused — the state is visible at a glance.
As always: speech recognition runs locally and offline by default; cloud backends are strictly your explicit choice. The app sends nothing but a heartbeat and the license check.
To update: the console will offer the update itself, or download the installer from the website.