{ lib , stdenv , buildPythonPackage , fetchurl , python , pytorch-bin , pythonOlder , pythonAtLeast }: buildPythonPackage rec { pname = "torchaudio"; version = "0.10.0"; format = "wheel"; src = let pyVerNoDot = lib.replaceStrings [ "." ] [ "" ] python.pythonVersion; unsupported = throw "Unsupported system"; srcs = (import ./binary-hashes.nix version)."${stdenv.system}-${pyVerNoDot}" or unsupported; in fetchurl srcs; disabled = ! (pythonAtLeast "3.7" && pythonOlder "3.10"); propagatedBuildInputs = [ pytorch-bin ]; # The wheel-binary is not stripped to avoid the error of `ImportError: libtorch_cuda_cpp.so: ELF load command address/offset not properly aligned.`. dontStrip = true; pythonImportsCheck = [ "torchaudio" ]; postFixup = '' # Note: after patchelf'ing, libcudart can still not be found. However, this should # not be an issue, because PyTorch is loaded before torchvision and brings # in the necessary symbols. patchelf --set-rpath "${lib.makeLibraryPath [ stdenv.cc.cc.lib ]}:${pytorch-bin}/${python.sitePackages}/torch/lib:" \ "$out/${python.sitePackages}/torchaudio/_torchaudio.so" ''; meta = with lib; { description = "PyTorch audio library"; homepage = "https://pytorch.org/"; changelog = "https://github.com/pytorch/audio/releases/tag/v${version}"; # Includes CUDA and Intel MKL, but redistributions of the binary are not limited. # https://docs.nvidia.com/cuda/eula/index.html # https://www.intel.com/content/www/us/en/developer/articles/license/onemkl-license-faq.html license = licenses.bsd3; platforms = platforms.linux; maintainers = with maintainers; [ junjihashimoto ]; }; }