Starstruk Logo MarkStarstruk

v1.4.7

modelscope/FunASRv1.4.7Aug 29, 2026by github-actions[bot]

AI Summary

This maintenance release addresses critical runtime decoding issues for SenseVoice and introduces performance profiling tools for real-time processing.

Key Highlights

  • Fix for SenseVoice F16 query embedding decoding.
  • Fix for SenseVoice weight uploads to selected backends.
  • Introduction of an AutoModel runtime adapter for Moss.
  • Addition of opt-in decode profiling for real-time performance tuning.

New Features

  • AutoModel runtime adapter for Moss.
  • Opt-in decode profiling tool.

Full Release Notes

## Python packages

The exact PyPI distributions built from tag commit `00dd1a54e077d0823263dfc4f6f01ef686b8fde9` are attached here alongside the nine self-contained llama.cpp runtime archives.

| Asset | SHA-256 |
|---|---|
| [funasr-1.4.7-py3-none-any.whl](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-1.4.7-py3-none-any.whl) | `7a0a87c0eadef8e83c3c53f0be5f6c23c96a85394b299dec413420a3852364dd` |
| [funasr-1.4.7.tar.gz](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-1.4.7.tar.gz) | `52263caa9ee13bc1ae8bb5bd334969bd77d74f4fd1bf211805a24472953c39dd` |

## What's Changed
* fix(runtime): decode SenseVoice F16 query embeddings by @LauraGPT in https://github.com/modelscope/FunASR/pull/3550
* docs(runtime): prepare llama.cpp v0.2.4 release by @LauraGPT in https://github.com/modelscope/FunASR/pull/3551
* docs(site): publish llama.cpp v0.2.4 evidence by @LauraGPT in https://github.com/modelscope/FunASR/pull/3552
* fix(runtime): upload SenseVoice weights to selected backend by @LauraGPT in https://github.com/modelscope/FunASR/pull/3555
* docs(ecosystem): add MOSS diarization deployment path by @LauraGPT in https://github.com/modelscope/FunASR/pull/3556
* docs(runtime): publish verified v0.2.5 download matrix by @LauraGPT in https://github.com/modelscope/FunASR/pull/3557
* feat(moss): add AutoModel runtime adapter by @LauraGPT in https://github.com/modelscope/FunASR/pull/3558
* docs(moss): expose AutoModel deployment contract by @LauraGPT in https://github.com/modelscope/FunASR/pull/3559
* test(release): track current llama.cpp runtime by @LauraGPT in https://github.com/modelscope/FunASR/pull/3560
* Fix tiny-gap semantic boundaries in readable SRT by @LauraGPT in https://github.com/modelscope/FunASR/pull/3562
* perf(realtime): add opt-in decode profiling by @LauraGPT in https://github.com/modelscope/FunASR/pull/3561
* docs(moss): sync AutoModel deployment matrices by @LauraGPT in https://github.com/modelscope/FunASR/pull/3563
* chore(release): prepare 1.4.7 by @LauraGPT in https://github.com/modelscope/FunASR/pull/3564


**Full Changelog**: https://github.com/modelscope/FunASR/compare/v1.4.6...v1.4.7

<!-- funasr-runtime-downloads:start -->
## Runtime downloads

This Python release pairs with the current prebuilt llama.cpp / GGUF runtime release: [runtime-llamacpp-v0.2.5](https://github.com/modelscope/FunASR/releases/tag/runtime-llamacpp-v0.2.5).

The same verified runtime assets are attached directly to this Python release so users can find the package and self-contained `llama-funasr-*` binaries in one place.

| Platform | Asset | SHA-256 |
|---|---|---|
| Linux arm64 | [funasr-llamacpp-linux-arm64.tar.gz](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-linux-arm64.tar.gz) | `4350ed49b1220374231ddafc84fed2dbe6a1d5e19d830edcb6a252d19d78b5db` |
| Linux x64 AVX2 | [funasr-llamacpp-linux-x64-avx2.tar.gz](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-linux-x64-avx2.tar.gz) | `f9bb3fe78776ef86feb334b02488d3b1a5c16926a9a1cc79b3d5b1219ef1fb30` |
| Linux x64 Vulkan | [funasr-llamacpp-linux-x64-vulkan.tar.gz](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-linux-x64-vulkan.tar.gz) | `88cd52ba1ce92af9bf344944521a9d47ed9ceb424b312a5993369228c1e8c4fb` |
| Linux x64 portable | [funasr-llamacpp-linux-x64.tar.gz](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-linux-x64.tar.gz) | `69d47bf724216093bedc89a50efe0a97c80a2f4b89367378a72d2d3d94e4f241` |
| macOS arm64 | [funasr-llamacpp-macos-arm64.tar.gz](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-macos-arm64.tar.gz) | `c6b3cea2c5238ae48175c931365231b1c47351f95ca289455536de0a959648af` |
| Windows x64 AVX2 | [funasr-llamacpp-windows-x64-avx2.zip](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-windows-x64-avx2.zip) | `6f19b8af0ce623122c767a0160bb8faabbc9666ae4f0cef1675cdf6250a9ea26` |
| Windows x64 CUDA | [funasr-llamacpp-windows-x64-cuda.zip](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-windows-x64-cuda.zip) | `3bb704e6b86f2e2e613aba84c361a51273e729c107183c7b69a1f2781c95cce8` |
| Windows x64 Vulkan | [funasr-llamacpp-windows-x64-vulkan.zip](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-windows-x64-vulkan.zip) | `d51e34afed66c0898dba3d944cb17d2abc92c5fb7758fa4ee376561d5ce35a4c` |
| Windows x64 portable | [funasr-llamacpp-windows-x64.zip](https://github.com/modelscope/FunASR/releases/download/v1.4.7/funasr-llamacpp-windows-x64.zip) | `dc90e85bc9a22477c5a8b556427fa5a6e0caf8951b5b6ec0237317406ce11bf5` |

Quick start: download one asset, unpack it, then run the bundled `download-funasr-model.sh <sensevoice|paraformer|nano>` helper and one of `llama-funasr-cli`, `llama-funasr-sensevoice`, or `llama-funasr-paraformer`.

For Python users, install from PyPI:

```bash
python -m pip install -U "funasr==1.4.7"
```
<!-- funasr-runtime-downloads:end -->