Changelog
What shipped, when, and what changed for the people using Dictivo every day.
v0.3.38 Public beta
Maintenance release.
- No changes to how dictation works. Nothing was added, removed, or moved.
- For people who turned on anonymous usage statistics, Dictivo now records whether microphone access was granted or denied. It is the most common reason setup fails, and until now it was the one thing we could not see. No audio, no transcripts, no app or file names - just granted or denied, once per install.
- If you never turned anonymous usage statistics on, this changes nothing: nothing was sent before and nothing is sent now. You can turn it off at any time in Settings.
- Updated the public Mac and Windows download to the latest Dictivo build.
v0.3.37Public beta
Maintenance release.
- No changes to how dictation works. Nothing was added, removed, or moved.
- Fixed a gap in the opt-in anonymous usage statistics: one onboarding milestone was dropped before you were asked for consent, so the onboarding funnel was reported incomplete. If you never opted in, nothing was sent before and nothing is sent now.
v0.3.36Public beta
Maintenance release.
- No changes to how dictation works.
- For people who turned on anonymous usage statistics, events now carry the app version they came from, so a problem introduced by one release can be told apart from a problem that was always there. If you never turned usage statistics on, nothing is sent.
v0.3.35Public beta
Model downloads you can watch, cancel, and resume.
- Local model downloads now show live progress and can be cancelled and resumed, both during onboarding and in the model manager. If a model will not fit the machine, Dictivo suggests a smaller one instead of failing late.
- Models download from Dictivo's own mirror first and fall back to the original source, so a slow or unavailable upstream no longer blocks local setup.
- Your license, trial state, and anonymous device ID are now also kept in the operating system credential store (Keychain on macOS, Credential Manager on Windows) and restored automatically after a reinstall or a wiped profile.
- Cloud Fast device sign-out is now self-serve from Settings, so freeing a device no longer needs a support request.
v0.3.34Public beta
Hotkey setup in onboarding, and a trial you can always see.
- Onboarding now includes a hotkey step with a real test dictation, plus recovery when the shortcut you picked is already taken by another app.
- A persistent alert appears if the dictation hotkey stops being available, instead of dictation silently doing nothing.
- The 14-day full Local trial now starts when you begin setup rather than at first launch, and the days remaining stay visible.
- Starting the trial sends a one-time anonymous activation ping (hashed device ID, platform, app version) so trials can be counted. It never includes audio, text, or personal data, and it is separate from the opt-in usage statistics setting. See the Privacy Policy.
v0.3.33Public beta
Up to 3x faster local dictation on Apple Silicon.
- Local transcription now runs on the Metal GPU on Apple Silicon Macs. Calibration measures both CPU and Metal and picks the faster engine - measured 1.9-2.9x faster end-to-end per model on an M4 Pro.
- The highest-quality local model (Large v3) is now comfortably interactive fully on-device: about 25 seconds per minute of audio on an M4 Pro, down from about 49.
- Settings and the dictation workbench show a live engine indicator (Metal GPU, Vulkan GPU, or CPU) in every display language, refreshed after each dictation so a GPU-to-CPU fallback is visible immediately.
- Measured results and method: Mac dictation benchmarks.
v0.3.32Public beta
Language display, sound themes, and release sync.
- Added display language options for German, French, Spanish, Italian, Dutch, Portuguese, Chinese, Japanese, and Korean while dictation language detection stays automatic.
- Refined dictation sound themes so start, stop, success, and error cues stay short and less fatiguing.
- Updated the public Mac download to the latest Dictivo build.
- Kept the website download and in-app version messaging aligned.
v0.3.13Public beta
Reliable dictation sounds and mute control.
- Start and stop cues now use native macOS playback first, keeping dictation sounds reliable over long sessions.
- The Settings preview buttons now play the selected cue, and the new Off option mutes dictation start and stop sounds.
- The latest desktop installer is available from the Dictivo website.
v0.3.12Public beta
One-click install and restart for updates.
- Check for updates now installs the downloaded update immediately and restarts Dictivo into the new version.
- The Settings and update banner copy now says install and restart, matching the actual update flow.
- The updater keeps existing settings, licenses, and local data in place during the app update.
- The latest desktop installer is available from the Dictivo website.
v0.3.11Public beta
Cloud Fast recovery for upgraded installs.
- Fixed upgraded Windows installs that could stay stuck on Local setup after Cloud Fast was activated.
- Manual Cloud Fast activation and cached Cloud Fast licenses now switch the desktop app into Cloud Fast when Local is not ready.
- The floating companion now distinguishes Local setup blocks from Cloud Fast subscription blocks.
- The latest desktop installer is available from the Dictivo website.
WindowsPublic beta
Windows x64 downloads and in-app updates are available.
- The Windows NSIS installer is now published with the same stable release version as macOS.
- Windows Check for updates reads the shared Dictivo updater manifest.
- The MSI installer is available for users who prefer that package format.
v0.2.8Public beta
Local first, Cloud Fast optional.
- Added the Local / Cloud Fast mode switch to the desktop dictation surface.
- Cloud Fast runs as a separate optional subscription for selected recordings.
- Settings now shows Local model controls only when Local mode is selected.
- Text cleanup is shared across Local and Cloud Fast.
- Floating companion behavior was updated for hotkey-driven recording and completion states.
- Native desktop recording keeps hotkey dictation working when the main window is minimized.
v0.2.0Public beta
Onboarding tiers & history.
- Onboarding matches the local engine tier to detected hardware on first launch.
- History view keeps recent dictations reachable for review and reuse.
- Dedicated panes for privacy controls and local engine settings.
- Companion overlay during dictation keeps the transcript visible while you work.
- Mac download path established ahead of later Windows packaging.
- Windows packaging groundwork added ahead of the public beta.
v0.1.0Private alpha
Initial dictation surface.
- Local-first dictation for turning speech into text on the desktop.
- Active transcript surface with a visible recording state.
- Hotkey scaffolding and a local user dictionary.
- macOS and Windows builds from the same workspace.