Customized build of 0xShug0/audio.cpp for Subwave — adds VibeVoice auto speaker-prefixing, cross-engine voice-alias resolution, and peak-normalized (non-clipping) audio output via a Python proxy layer. cuda12 = CUDA backend (official base image + proxy), vulkan = built from source for AMD/Vulkan (no official Vulkan image exists upstream). Not affiliated with the original project.
Content type
Image
Digest
sha256:1f6e59217…
Size
2.7 GB
Last updated
14 days ago
docker pull k8bit/audiocpp-proxy:cuda12