Repository navigation
Tags
Tags
v1.6.10 Speaker diarization controls, Integrations view, dedicated Meetings settings, cloud-sync tombstones for folders and transcriptions, and a settings reorganization that collapses AI Models into Speech-to-Text and Language Models. Includes post-release fixes: - Cloud-api error surfacing (cloud request errors now include message/code/stack) - macOS app startup packaging restoration - Batch transcription tombstone sync (single request per 100 ids instead of per-row loop) See CHANGELOG.md for full details.
Merge pull request OpenWhispr#582 from OpenWhispr/feat/speaker-diariz… …ation feat: speaker diarization + meeting AEC
chore: update changelog/readme for v1.6.6 and regenerate lockfile wit… …h node 22 Add missing changelog entries for native macOS audio tap, KDE Wayland shortcuts, new local models, permission simplification, and multiple bug fixes. Add KDE Wayland to README features. Regenerate package-lock.json with Node 22.
fix: force x64 ffmpeg download for Intel Mac builds ffmpeg-static's install script skips download if the binary already exists, regardless of architecture. On arm64 CI runners, npm ci installs the arm64 ffmpeg first, then the x64 reinstall is silently skipped. This caused Intel Mac builds to ship with an arm64 ffmpeg, breaking local transcription with EBADARCH (error -86). Delete the existing binary before re-running the install to force a fresh x64 download.
fix: regenerate lockfile with full npm install for cross-platform deps The previous lockfile was generated with --package-lock-only which omits optional platform-specific dependencies (known npm bug). This caused rollup native binaries (@rollup/rollup-win32-x64-msvc, @rollup/rollup-linux-x64-gnu) to be missing, breaking CI builds on Windows and Linux.
PreviousNext