v0.9.15
Fixed
- Correctly-detected languages no longer dropped for a missing mapping. The ISO 639-1→639-2 table only had ~50 common languages, so detections like
nn(Nynorsk) were discarded. Expanded to faster-whisper's full set, with a babelfish fallback for anything else. - More languages recognized from track titles. Added ~45 languages (Wolof, Swahili, Tamil, Welsh, …) to title detection — fixes
undtracks whose title named a language the map didn't cover. - Forced subtitles with late dialogue now detect. Embedded text-sub extraction no longer caps at the first 10 minutes, so a forced sub whose lines start later isn't extracted empty.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.15 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.15-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.15-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.9.15-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.