onnx-asr v0.12.0

Highlights

• Added support for GigaAM Multilingual CTC and GigaAM Multilingual Large CTC.
• Added convolution-based ONNX preprocessors, providing much faster preprocessing with the CUDAExecutionProvider.
• Current onnxruntime-gpu versions require CUDA 13. See the installation guide for CUDA 12 and CUDA 13 installation commands.

What's Changed

• Bump requests from 2.32.5 to 2.33.0 in the uv group across 1 directory by @dependabot[bot] in #109
• Bump the update-actions group with 2 updates by @dependabot[bot] in #114
• Support loading ORT models as a fallback by @mscdex in #118
• Update deps by @istupakov in #120
• Fix compatible onnxruntime versions, update deps by @istupakov in #123
• Bump the update-actions group across 1 directory with 3 updates by @dependabot[bot] in #127
• Bump actions/cache from 5 to 6 in the update-actions group by @dependabot[bot] in #128
• Bump ghcr.io/devcontainers/features/common-utils from 2.5.8 to 2.5.9 by @dependabot[bot] in #129
• Add Conv-based STFT variants for the ONNX preprocessors by @intexcor in #124
• Update CUDA by @istupakov in #130
• Fix zero-length segments after VAD by @istupakov in #132
• Add new GigaAM Multilingual models by @istupakov in #133
• Little fixes by @istupakov in #134

New Contributors

• @mscdex made their first contribution in #118
• @intexcor made their first contribution in #124

Full Changelog: v0.11.0...v0.12.0