Fun-ASR-Nano-2512 is an end-to-end automatic speech recognition (ASR) model. For authoritative capabilities, supported languages, and constraints, see the upstream model card: https://huggingface.co/FunAudioLLM/Fun-ASR-Nano-2512

Export provenance (how these ONNX files were obtained)

The ONNX files in this repo were downloaded from ModelScope: https://www.modelscope.cn/models/zengshuishui/FunASR-nano-onnx/files

The export script referenced for producing the ONNX package is: https://github.com/Wasser1462/FunASR-nano-onnx

Author: https://github.com/Wasser1462

License

Upstream model license is Apache-2.0 (see upstream model card / discussions): https://huggingface.co/FunAudioLLM/Fun-ASR-Nano-2512

This repo redistributes exported artifacts; please also comply with any upstream notices.

Intended use

Offline / local ASR with sherpa-onnx using ONNX runtime compatible backends.
Use-cases: transcription, speech-to-text pipelines.

Limitations / Notes

Accuracy and behavior depend on decoding settings, audio preprocessing, and runtime.
This is an ONNX export; results may differ from upstream Python inference.

Citation

If you use Fun-ASR in research, cite the upstream technical report:

bibtex

@article{an2025fun,
  title={Fun-ASR Technical Report},
  author={An, Keyu and Chen, Yanni and Deng, Chong and Gao, Changfeng and Gao, Zhifu and Gong, Bo and Li, Xiangang and Li, Yabin and Lv, Xiang and Ji, Yunjie and others},
  journal={arXiv preprint arXiv:2509.12508},
  year={2025}
}

References

Upstream model: https://huggingface.co/FunAudioLLM/Fun-ASR-Nano-2512
Upstream project: https://github.com/FunAudioLLM/Fun-ASR
ModelScope export source: https://www.modelscope.cn/models/zengshuishui/FunASR-nano-onnx/files
Export script: https://github.com/Wasser1462/FunASR-nano-onnx

aoiandroid/sherpa-onnx-funasr-nano-int8-2025-12-30

sherpa-onnx-funasr-nano-int8-2025-12-30 (ONNX export)

Summary

What is Fun-ASR-Nano-2512?