Repository navigation
Sorla 1.0.2
Bug fixes, accessibility, calmer feedback without notifications, one place for updates, and more privacy. Full list below.
- Updates for Sorla and the speech model live in one place, Settings › Updates: a row for each with its version and result, one Check Now that asks GitHub and Hugging Face together, and two toggles (check automatically, install automatically) that replace the model-only ones and keep their values. Check for Updates… in the menu opens that section and runs the check, a newer Sorla shows in the menu's status row, and the About window is back to just the version.
- Sorla no longer uses notifications, so macOS never asks for a third permission. Problems show in the indicator (an hourglass while the model gets ready, a warning when something failed), VoiceOver announces them, and the menu's status row explains the last dictation (muted microphone, text on the clipboard, no microphone, failed transcription) until the next one starts or for five minutes.
- The update check tells "offline", "GitHub rate limit" and "bad response" apart and never shows a failure as "up to date".
- About window: calmer layout with the Klang Pianissimo credit up front, links to the website, GitHub and privacy page; other credits and the license texts are in Credits and Licenses….
- When the model couldn't be loaded, dictation is refused up front with a pointer to Try Again in the menu.
- Cancelling a recording while an earlier one is still being transcribed no longer hides the indicator early or lets a model update swap in mid-transcription.
- Something you copy between two quick dictations is kept instead of being replaced by the older clipboard content.
- A modifier trigger only stops recording once the key is really up, and Sorla's own paste can't read as a release.
- A dictation that gives no text now says why: the indicator briefly shows a symbol (nothing heard, no text, muted microphone, text on the clipboard) and VoiceOver announces it; a muted microphone also shows while recording.
- Keyboard: Esc and ⌘W close the Settings, About and Welcome windows, ⌘Q quits from any of them, and the Try it here field gets focus as soon as it appears.
- The first menu item starts and stops dictation (Start Dictation (Hold Right ⌘) / Stop Dictation), so Voice Control, Switch Control and head-pointer users can dictate without the trigger key. The text lands in the app you were working in, and a press and release of the key also stops a dictation started from the menu (a ⌘-shortcut doesn't).
- With VoiceOver on, Sorla says "Pasted" once the text has been pasted, after the recording has ended, so the microphone never hears it.
- Welcome window with VoiceOver: the app icon and row symbols are no longer read out, a finished row says "Done", a busy one "In progress", each row is read as one group, and VoiceOver announces when Sorla is ready.
- The menu bar icon shows an exclamation mark in place of the dot when the model couldn't be loaded, and VoiceOver reads it as "Sorla (model couldn't be loaded)"; while transcribing it reads "Sorla (transcribing)".
- The Welcome window and the help page describe a finished row by its checkmark instead of its colour ("shows a checkmark", not "turns green").
- Warnings and hints in Settings and the Welcome window (the Fn key warning, the Login Items approval hint, a failed update's reason, the Accessibility restart hint) use a larger size and the primary text colour, so they stay readable with Increase Contrast.
- Settings with VoiceOver: Check Now, Download and Try Again say what they act on ("Check for updates now", "Try downloading the model again"), and the Model row is plain text instead of a picker with a single choice.
- The last transcription is kept for Paste Last for five minutes, not until Sorla quits, and is forgotten at once when the screen locks, the Mac sleeps or you switch user. Paste Last is then unavailable until the next dictation. Locking, sleeping or switching user also cancels a recording in progress and drops a transcription that hasn't been pasted yet.
- A recording stops by itself after five minutes and what was said is transcribed and pasted as usual, so a forgotten toggle recording doesn't keep the microphone open. The indicator's bars dim during the last ten seconds.
- With automatic checks on, the speech model is checked about once a day, like Sorla itself, instead of at every launch.
- Start Dictation chosen while Settings or About is in front pastes into the app you were using before, like Paste Last.
- Closing the last Sorla window returns you to the app you were using, so a following ⌘Q doesn't quit Sorla by accident.
- With VoiceOver on, Settings notes under Play sounds that the sounds tell you when recording starts and stops.
- The two update toggles take their values from 1.0.1's model-only settings (autoCheckModelUpdates, autoDownloadModelUpdates), which are then removed; going back to 1.0.1 afterwards shows its automatic model checks and downloads as off.
Update
Choose Check for Updates… in Sorla's menu. To update: quit Sorla, download the DMG below, drag Sorla to Applications and choose Replace. Microphone and Accessibility permissions are kept. Your update settings from 1.0.1 carry over.
Install
Download Sorla-1.0.2.dmg below, open it and drag Sorla to Applications. Or with Homebrew: brew install --cask markstrom/tap/sorla. Requires a Mac with Apple Silicon and macOS 14 or later. Signed with Developer ID and notarized by Apple.
Speech recognition by Klang Pianissimo from Klang AI AB (CC BY 4.0). Sorla is an independent app and is not made, endorsed or supported by Klang AI AB.
SHA-256 (Sorla-1.0.2.dmg): 15ed0a25c038ffe87045c1986f046fd0600799eedcb1f381a9a72d4747e5bed4