karaoke-assets

Runtime assets for a personal browser karaoke app. The app downloads these files from this repo at run time, checks each one against a pinned SHA-256, and runs them locally in the browser with ONNX Runtime Web.

Not for training. These files are hosted only so the app can load them; they are not a dataset or a model release.

Files

Path	What it is	Upstream and licence
vendor/ort/ort-wasm-simd-threaded.asyncify.wasm	ONNX Runtime Web WASM binary (onnxruntime-web 1.31.1)	onnxruntime-web, MIT
models/mdx_voc_ft_fp16.onnx	UVR MDX-Net Voc_FT vocal separation; BatchNorm folded, fp16 weights and compute (fp32 I/O)	Derived from UVR MDX-Net Voc_FT, from the TRvlvr/model_repo release all_public_uvr_models. The UVR application is MIT (Anjok07/ultimatevocalremovergui), but that release carries no licence file, so redistribution terms are unverified.
models/mdx_voc_ft_fp32.onnx	Same model, fp32, Conv+BatchNorm pairs folded by ONNX Runtime	Same as above: redistribution terms unverified.
models/fcpe.onnx	FCPE pitch estimator	MIT (CNChTu / torchfcpe, Copyright (c) 2023 CN_ChiTu). ONNX export by Scicom-intl (MIT), from Scicom-intl/fcpe-onnx.
tools/binaryen-version_133-x86_64-linux.tar.gz	binaryen version_133 (wasm-opt), mirrored byte-for-byte from the WebAssembly/binaryen GitHub release, only as a CI fallback for when GitHub release downloads fail	binaryen, Apache-2.0
tools/tailwindcss-v4.3.5-linux-x64	Tailwind CSS standalone CLI v4.3.5, mirrored byte-for-byte from the tailwindlabs/tailwindcss GitHub release, only as a CI fallback	tailwindcss, MIT

If you are a rights holder and want a file removed, open a discussion on this repo.