sci-ml/torchaudio
Data manipulation and transformation for audio signal processing
Screenshots
-
torchaudio-2.11.0-r1~amd64cuda rocm ffmpeg openmp +rnnt test python_single_target_python3_12 python_single_target_python3_13 python_single_target_python3_14 debug
View 
Download 
Browse License: BSD Overlay: phackerlay
ChangeLog
commit bf36b5de1387a7a12491744344186e3e183f1a77
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed Aug 12 10:34:53 2026 +0200
sci-ml/torchaudio: add 2.11.0-r1 against pytorch-2.13
torchaudio 2.11.0 is upstream's latest (pytorch/audio lags torch), so
there is no distinct 2.13 release. Ship the same source as a second
revision pinned =pytorch-2.13* (r1) alongside r0's =pytorch-2.11*, so
both the torch-2.11 (vllm-0.26.0) and torch-2.13 (vllm-0.27.1) frontends
have a matching torchaudio; consumers' ~torchaudio-2.11.0 backtracks to
the right rev.
commit 5a2146660cab45fafc395991781040c9176065ce
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Jun 18 00:04:58 2026 +0200
sci-ml/torchaudio: Keyword 2.11.0 for ~arm64
Torch C++/ffmpeg audio I/O extension; builds against sci-ml/pytorch on
aarch64 and imports cleanly. C++ (CMake); aarch64-portable.
commit 0c7a37e26c09ea8a656e150b923d5df190254185
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 13 14:35:43 2026 +0200
sci-ml/torchaudio: disable py3.11
commit 6bcf3b622b30ccfb056615888e41c30e1a8a8bc4
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu May 7 08:49:23 2026 +0200
sci-ml/torchaudio: new package, 2.11.0
PyTorch audio ops, sister of sci-ml/torchvision. Required by
dev-python/vllm in the cpu path.
CPU-only build for now: USE_CUDA=0 / USE_ROCM=0 /
BUILD_CUDA_CTC_DECODER=0. CUDA/ROCm modes need the same setup_helpers
env-var dance torchvision does — defer until a USE-flag-driven cycle.
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed Aug 12 10:34:53 2026 +0200
sci-ml/torchaudio: add 2.11.0-r1 against pytorch-2.13
torchaudio 2.11.0 is upstream's latest (pytorch/audio lags torch), so
there is no distinct 2.13 release. Ship the same source as a second
revision pinned =pytorch-2.13* (r1) alongside r0's =pytorch-2.11*, so
both the torch-2.11 (vllm-0.26.0) and torch-2.13 (vllm-0.27.1) frontends
have a matching torchaudio; consumers' ~torchaudio-2.11.0 backtracks to
the right rev.
commit 5a2146660cab45fafc395991781040c9176065ce
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu Jun 18 00:04:58 2026 +0200
sci-ml/torchaudio: Keyword 2.11.0 for ~arm64
Torch C++/ffmpeg audio I/O extension; builds against sci-ml/pytorch on
aarch64 and imports cleanly. C++ (CMake); aarch64-portable.
commit 0c7a37e26c09ea8a656e150b923d5df190254185
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Wed May 13 14:35:43 2026 +0200
sci-ml/torchaudio: disable py3.11
commit 6bcf3b622b30ccfb056615888e41c30e1a8a8bc4
Author: Ivan S. Titov <iohann.s.titov@gmail.com>
Date: Thu May 7 08:49:23 2026 +0200
sci-ml/torchaudio: new package, 2.11.0
PyTorch audio ops, sister of sci-ml/torchvision. Required by
dev-python/vllm in the cpu path.
CPU-only build for now: USE_CUDA=0 / USE_ROCM=0 /
BUILD_CUDA_CTC_DECODER=0. CUDA/ROCm modes need the same setup_helpers
env-var dance torchvision does — defer until a USE-flag-driven cycle.