New features, improvements, and fixes in each version of Yawn.
New AI note generation is retired. Your notes, saved drafts, and transcripts remain available.
Improved
- After Stop, Yawn clearly separates recording from processing saved audio. Transcription no longer leaves an active Stop control on screen.
- Existing AI notes are labeled Saved AI draft. A failed or interrupted draft no longer promises generation or offers a retry.
Fixed
- Restored saved-meeting access when macOS rejects the previous network-denial profile used by the local reader.
- Opening the meetings folder in Finder no longer lets its .DS_Store metadata file block meeting recovery.
Before upgrading
- New AI note generation, regeneration, and note-model downloads are unavailable in this release.
- This update preserves personal notes, transcripts, saved AI drafts, and existing model files. It does not delete meeting data or reclaim model storage.
- Yawn requires an Apple-silicon Mac running macOS 14.4 or later. Apple speech requires macOS 26.
Known limitations
- Yawn remains an internal alpha. Review saved AI drafts against their linked transcript passages before relying on them.
- Optional speaker suggestions cover retained microphone audio only. They are anonymous review hints and can be wrong around short turns or overlap.
- Searching across transcript contents remains experimental and disabled by default.
Optional speaker suggestions for finished in-person recordings.
Added
- Download the separate 107 MB speaker-analysis model in Settings, then choose Analyze speakers on a finished meeting with a retained microphone recording and transcript.
- Review anonymous speaker clusters and possible overlap beside microphone transcript turns. Yawn does not infer names or rewrite the transcript.
Improved
- When recording access blocks Record in a selected meeting, the relevant permission action appears beside it.
- Settings explanations use larger text, and long meeting titles fit better with the inspector open.
Before upgrading
- Yawn requires an Apple-silicon Mac running macOS 14.4 or later. Apple speech requires macOS 26.
- Speaker analysis is optional and local. It needs the meeting microphone recording to remain retained.
Known limitations
- Speaker suggestions can be wrong around short turns and overlap. They are review hints, not word-level attribution or a person identity.
- Analysis covers only the microphone recording. It does not separate people on the system-audio leg or replace existing Me and Them labels.
- Yawn remains an internal alpha. Generated meeting notes are drafts; review their linked transcript passages before relying on them.
Start recording without opening Settings after every launch.
Improved
- When microphone access is already allowed, Yawn measures system-audio access once after launch so Record can unlock without a separate Settings visit.
- If system audio still needs an explicit allow, that control appears beside the disabled Record button — including when a meeting is selected.
Fixed
- Copy no longer sends you to Settings only to enable system audio when the allow action is available in the main window.
Before upgrading
- Yawn requires an Apple-silicon Mac running macOS 14.4 or later. Apple speech requires macOS 26.
- Speech and note models are separate optional downloads. Existing downloaded-model choices are preserved.
Known limitations
- Yawn remains an internal alpha. Generated meeting notes are drafts; review the linked transcript passages before relying on them.
- macOS has no status API for system audio, so each app launch still performs a disposable check. That check is automatic when the microphone is already authorized.
- Searching across transcript contents is experimental and disabled by default.
Easier first setup, clearer Settings, and a softer dark appearance.
Added
- Use Apple’s on-device speech engine on macOS 26, with language files managed by macOS.
Improved
- Start with Apple speech when it is ready, or choose a downloaded speech model. You can add a larger model later in Settings.
- Your selected speech engine stays selected after restarting Yawn, and saved transcripts stay available when you change engines.
- Settings groups recording access, transcription, meeting notes, and storage so the next action is easier to find.
- Long titles, transcript controls, source previews, and Trash layouts wrap and scroll within supported window sizes.
- Lavender Haze gives dark mode softer charcoal backgrounds, muted lavender accents, and gentler text colors.
Fixed
- A microphone that stops delivering audio during a recording now ends that take with a microphone-specific message instead of marking it complete.
Before upgrading
- Yawn requires an Apple-silicon Mac running macOS 14.4 or later. Apple speech requires macOS 26.
- Speech and note models are separate optional downloads. Existing downloaded-model choices are preserved.
Known limitations
- Yawn remains an internal alpha. Generated meeting notes are drafts; review the linked transcript passages before relying on them.
- Searching across transcript contents is experimental and disabled by default. It is not generally available in 0.6.4.
Switching between meetings is faster.
Improved
- Opening a saved meeting, its transcript, or its note takes less time in an existing library.
- In six same-Mac checks with an eight-meeting library, switching fell from 1.30–1.84 seconds in 0.6.2 to 195–217 milliseconds in 0.6.3.
Known limitations
- Switching time varies with your Mac and meeting library. The measured range came from one eight-meeting library on one Mac.