Date: Thu, 14 Dec 2023 02:13:28 GMT From: Yuri Victorovich <yuri@FreeBSD.org> To: ports-committers@FreeBSD.org, dev-commits-ports-all@FreeBSD.org, dev-commits-ports-main@FreeBSD.org Subject: git: b0297e87d00d - main - audio/py-torchaudio: New port: PyTorch-based audio signal processing and machine learning library Message-ID: <202312140213.3BE2DSkZ009141@gitrepo.freebsd.org>
next in thread | raw e-mail | index | archive | help
The branch main has been updated by yuri: URL: https://cgit.FreeBSD.org/ports/commit/?id=b0297e87d00d8174ef39845e5faba0a4e1c8de03 commit b0297e87d00d8174ef39845e5faba0a4e1c8de03 Author: Yuri Victorovich <yuri@FreeBSD.org> AuthorDate: 2023-12-14 02:12:18 +0000 Commit: Yuri Victorovich <yuri@FreeBSD.org> CommitDate: 2023-12-14 02:13:24 +0000 audio/py-torchaudio: New port: PyTorch-based audio signal processing and machine learning library --- audio/Makefile | 1 + audio/py-torchaudio/Makefile | 46 +++++++++++++++++++++++++++++++++++++++++++ audio/py-torchaudio/distinfo | 5 +++++ audio/py-torchaudio/pkg-descr | 8 ++++++++ 4 files changed, 60 insertions(+) diff --git a/audio/Makefile b/audio/Makefile index f06f538b2b74..446c1cd6d3fc 100644 --- a/audio/Makefile +++ b/audio/Makefile @@ -667,6 +667,7 @@ SUBDIR += py-speechrecognition SUBDIR += py-spotipy SUBDIR += py-tagpy + SUBDIR += py-torchaudio SUBDIR += py-wavio SUBDIR += py-webrtcvad SUBDIR += py-zita-audiotools diff --git a/audio/py-torchaudio/Makefile b/audio/py-torchaudio/Makefile new file mode 100644 index 000000000000..2f623ef56c63 --- /dev/null +++ b/audio/py-torchaudio/Makefile @@ -0,0 +1,46 @@ +PORTNAME= torchaudio +DISTVERSIONPREFIX= v +DISTVERSION= 2.1.1 +CATEGORIES= audio python # machine-learning +MASTER_SITES= https://downloads.sourceforge.net/project/sox/sox/14.4.2/:sox +PKGNAMEPREFIX= ${PYTHON_PKGNAMEPREFIX} +DISTFILES= sox-14.4.2.tar.bz2:sox +EXTRACT_ONLY= ${GH_ACCOUNT}-${GH_PROJECT}-${DISTVERSIONFULL}_GH0.tar.gz + +MAINTAINER= yuri@FreeBSD.org +COMMENT= PyTorch-based audio signal processing and machine learning library +WWW= https://pytorch.org/audio/stable/index.html + +LICENSE= BSD2CLAUSE +LICENSE_FILE= ${WRKSRC}/LICENSE + +BUILD_DEPENDS= ${PYTHON_PKGNAMEPREFIX}pytorch>0:misc/py-pytorch@${PY_FLAVOR} \ + ${PYTHON_PKGNAMEPREFIX}wheel>0:devel/py-wheel@${PY_FLAVOR} \ + cmake:devel/cmake-core \ + ninja:devel/ninja \ + pybind11>0:devel/pybind11 +LIB_DEPENDS= libprotobuf.so:devel/protobuf +RUN_DEPENDS= ${PYTHON_PKGNAMEPREFIX}pytorch>0:misc/py-pytorch@${PY_FLAVOR} + +USES= python +USE_PYTHON= pep517 autoplist + +MAKE_ENV= USE_FFMPEG=0 # ffmpeg integration is broken or unavailable, see https://github.com/pytorch/audio/issues/3716 + +USE_GITHUB= yes +GH_ACCOUNT= pytorch +GH_PROJECT= audio + +post-patch: + @${REINPLACE_CMD} \ + -e 's| URL .*| URL file://${DISTDIR}/sox-14.4.2.tar.bz2|' \ + ${WRKSRC}/third_party/sox/CMakeLists.txt + +post-install: + @cd ${STAGEDIR}${PYTHONPREFIX_SITELIBDIR} && ${STRIP_CMD} \ + torchaudio/lib/_torchaudio.so \ + torchaudio/lib/_torchaudio_sox.so \ + torchaudio/lib/libtorchaudio.so \ + torchaudio/lib/libtorchaudio_sox.so + +.include <bsd.port.mk> diff --git a/audio/py-torchaudio/distinfo b/audio/py-torchaudio/distinfo new file mode 100644 index 000000000000..25f430b308a7 --- /dev/null +++ b/audio/py-torchaudio/distinfo @@ -0,0 +1,5 @@ +TIMESTAMP = 1702517865 +SHA256 (sox-14.4.2.tar.bz2) = 81a6956d4330e75b5827316e44ae381e6f1e8928003c6aa45896da9041ea149c +SIZE (sox-14.4.2.tar.bz2) = 935449 +SHA256 (pytorch-audio-v2.1.1_GH0.tar.gz) = fc905d8d80cec5060445f07314b5b4d276759636de87a4c21821d31d760c3974 +SIZE (pytorch-audio-v2.1.1_GH0.tar.gz) = 4414068 diff --git a/audio/py-torchaudio/pkg-descr b/audio/py-torchaudio/pkg-descr new file mode 100644 index 000000000000..67052afd9692 --- /dev/null +++ b/audio/py-torchaudio/pkg-descr @@ -0,0 +1,8 @@ +The aim of torchaudio is to apply PyTorch to the audio domain. By supporting +PyTorch, torchaudio follows the same philosophy of providing strong GPU +acceleration, having a focus on trainable features through the autograd system, +and having consistent style (tensor names and dimension names). Therefore, it is +primarily a machine learning library and not a general signal processing +library. The benefits of PyTorch can be seen in torchaudio through having all +the computations be through PyTorch operations which makes it easy to use and +feel like a natural extension.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?202312140213.3BE2DSkZ009141>