Dictly 1.2.1

Small quality-of-life release driven by user feedback in #6.

New: "Append a space after inserted text" (off by default)
Push-to-talk users asked for it: with the toggle on (Settings → Behaviour), every inserted phrase ends with a single space, so consecutive dictations chain naturally without reaching for the spacebar between them. Off by default — nothing changes unless you opt in.

Fixed: Dock icon no longer lingers
Opening Settings (or About) put Dictly's icon in the Dock and closing the window didn't remove it. The icon now disappears as soon as the last Dictly window is gone — including the About panel, a minimized Settings window, and the onboarding window.

Permissions at a glance
Settings → Permissions now shows live status for both permissions Dictly needs — Microphone and Accessibility — with a button that either fires the system prompt (if never asked) or jumps straight to the right System Settings pane. No need to reopen the onboarding window just to check.

Dictly 1.2

Two changes, both aimed at transcription quality out of the box.

Bundled model upgraded: Whisper Large v3 Turbo (compressed)
The app now ships with large-v3-v20240930 (547 MB, quantized) instead of base — near-Large-v3 accuracy for all languages with ~0.15 RTF on Apple Silicon. This is the model the app used to recommend as a separate download; now it works offline from the first launch. If you never picked a model manually, Dictly switches to it automatically. (The download is bigger for the same reason — ~500 MB.)

New: “Keep microphone warm for instant start” (on by default)
macOS powers the microphone hardware down when idle — the next capture start can take a couple of seconds, which is why the first word of a dictation sometimes went missing. Dictly now keeps the mic stream open between dictations: recording starts instantly, and up to half a second captured before you press the hotkey is included, so first words are never clipped.

• macOS shows the microphone indicator while Dictly runs — that's the honest cost. Turn the option off in Settings if you prefer.
• Bluetooth microphones are excluded (holding a capture stream open would pin them to low-quality call audio); they keep the on-demand start.

Dictly 1.1

Language, microphone and hotkey fixes — the two most-reported issues addressed.

Language (#1, #4)

• The default spoken language now follows your Mac's system language instead of being hard-coded (#4). First-run onboarding also asks for it explicitly.
• New: switch between two languages with a dedicated hotkey (#1). Enable it in Settings → “Switch between two languages with a hotkey”; the active language (flag or “auto”) shows next to the menu-bar icon and applies to your next dictation.

Microphone & audio

• New Microphone picker in the menu bar. Picking a mic sets it as the Mac's default input, and the recorder follows.
• Fixed repeated recordings capturing no audio on Bluetooth / Continuity / virtual microphones.

Hotkeys & permissions

• Fixed a bug where a second hotkey could never fire.
• The push-to-talk and language-switch hotkeys can't be assigned to the same key.
• Clearer Accessibility guidance: an actionable alert when auto-paste is blocked, a warning at launch when a modifier-only hotkey (Fn, right ⌥) can't work without the grant — and hotkeys now recover automatically the moment access is granted, no relaunch needed.
• The onboarding window stays fully on screen on small displays.

Known limitation: the bundled model is still Whisper base (~139 MB), which is weak for non-English speech. For best accuracy, download a larger model right after install in Settings → Transcription models — recommended: Large v3 · compressed (547 MB, one-time download).

Dictly 1.0

First public release.

• Whisper base multilingual bundled (~139 MB) — works offline from first launch
• For best accuracy, switch to a larger model right after install in
  Settings → Transcription models. Recommended: Large v3 · compressed
  (547 MB, one-time download from HuggingFace) — same quality as full
  Large v3 at a third of the size. RTF ≈ 0.15 on Apple Silicon (a 10-second
  utterance transcribes in ~1.5 seconds).
• 99 languages, fully on-device
• Push-to-talk hotkey (default: Fn)
• Auto-paste into the focused app via Accessibility