Fixed
- Audio detection can no longer hang the whole app. A specific track wedged whisper's transcribe, which (defaulting to all CPU cores) starved the event loop until a manual restart. Now: whisper's CPU threads are capped (leaving cores for the app), each transcribe is bounded by a timeout that abandons the track instead of retrying more windows, and the clip-extract ffmpeg is killed on timeout instead of orphaned. A pathological track is skipped (stays
und) and the app stays responsive.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.19 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.19-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.19-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.19-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.