Anarlog 1.4.14 keeps live meetings going through brief network drops, lets you pick local Whisper models and speaker devices, and makes sharing, sync, and speaker names more reliable.
Meetings and recording
Keep recording through a brief transcription or network drop instead of
ending the meeting, then reconnect with a short delay instead of retrying
immediately
Pick a speaker output in Settings → Meetings so loopback capture follows
the device you are actually using on Windows and Linux
Label the other person in a 1:1 transcript, and name speakers in larger
meetings when a unique high-confidence voiceprint match is available
Quit Completely from the tray or Dock confirmation without waiting for a
background flush
Transcription and intelligence
Choose a local whisper.cpp .bin model file from Settings → Transcription
instead of relying only on bundled downloads
Keep Mistral batch transcription speaker labels aligned with the words they
belong to
Load ChatGPT subscription models from the current catalog instead of a
stale fallback list
Sharing and teams
Set a default for new meeting-note shares in Settings → Meetings: Only
me, People in the meeting, or Everyone in your workspace
Email teammates when you invite them to a workspace, with a link they can
accept after signing in
Enforce Require SSO for claimed company domains so Google, GitHub, and
email sign-in cannot bypass workspace policy
Show Not invited on suggested attendees instead of a long explanation,
and keep Invite as the action that actually sends access
Sync
Rename this computer in Settings → Sync so it is recognizable on your other
devices
See restored notes appear while a new device is still finishing its first
sync
Read a clear sync-failure message that your notes stay safe on this device,
instead of a raw backend error
Account and contacts
Find Sign out and Sign out everywhere on the Account tab, not under
Developers
Create a contact from Enhance contact when the participant is not already
in your address book
Developers
Propose a summary or memo replacement from the CLI or MCP, then apply or
decline it in the desktop app
Linux
Keep the app running when the audio device glitches instead of crashing on
an ALSA stream error
