Releases: zakattack02/Whisper-Script
v3.0.0.2
v3.0.0.1
libnccl.so.2 added
v3.0.0.0
CUDA GPU support + semantic versioning restructure
- Dual CPU+GPU binaries (auto-selects CUDA when GPU is enabled)
- Bundles libcudart.so.12, libcublas.so.12, libcublasLt.so.12
- CPU-only zip available as separate asset (3.5 MB)
- Semantic versioning: all releases now use MAJOR.MINOR.PATCH.REVISION
See v0.0.0.98 for the original release of this code.
v2.2.0.0
Per-video sub-progress reporting. The scheduled task now reports progress as each chunk finishes, so the percentage bar updates smoothly instead of freezing per video.
v2.1.0.0
Fixed ffprobe path detection. Added FfprobePath config field + FindFfprobe() fallback chain + configPage.html input.
v2.0.0.0
Split long audio into 30-minute chunks to prevent OOM killer on low-RAM servers (4GB). Process each chunk separately and merge SRTs.
v1.1.2.0
Fixed -ngl 999 -> -dev 0 for GPU path (whisper-cli doesn't support -ngl). CPU path already fixed in 1.1.0.0/1.1.1.0.
v1.1.1.0
Fixed AVX-512 SIGILL: added GGML_NATIVE=OFF to cmake so binary doesn't include AVX-512 instructions from build machine (Ryzen 9950X3D)
v1.1.0.0
Fixed whisper-cli arguments: removed unsupported --output-dir and -vv flags, added -ng for CPU-only mode
v1.0.0.0
Add FFmpeg audio extraction for MP4/MKV support, polished card-based config page, fix PluginPageInfo for Jellyfin 10.11.2