You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
PDF viewer major upgrade: persistent per-PDF state (progress, zoom/rotation, highlights/marks, bookmarks, notes), “Resume reading”/clear-data actions, a new Reading Log dock, faster jumps in large PDFs, and improved click-to-read/text grouping (incl. better two-column handling and stable highlighting).
Smarter PDF/TTS workflow: automatic language detection + voice selection on PDF open; improved text normalization (e.g., hyphenated wraps); new Chatterbox ONNX TTS engine with prompt recording plus CoreML stability + fallback retries.
Audio improvements: open audio-only files, better segment selection/export, and cleaner resource/state teardown when switching content.
Tracking fix: correct global frame index mapping for CoTracker online inference (prevents “stuck” outputs) with regression tests.
Lean(er) packaging & deps: removed legacy YOLO/YOLOACT/YOLACT/SCOPS submodules/references, made heavy ML deps more optional (incl. PyTorch), and reduced PyInstaller bundle size.
Build/CI/docs polish: hardened release + PyPI publishing workflows, updated CI lint/docs publishing gates, refreshed Jupyter Book/docs, and added/updated tests across updated components.
