UL-UNAS

🎉 This is the official implementation of our IEEE TASLP paper:

UL-UNAS: Ultra-Lightweight U-Nets for Real-Time Speech Enhancement via Network Architecture Search

🔥 News

• [2026-2-13] Added streaming inference code ulunas_onnx implemented by Kailai Shen.
• [2026-2-4] The manuscipt is accepted by IEEE TASLP.
• [2026-2-3] The updated paper is uploaded to arxiv.
• [2026-2-1] The pre-trained checkpoint is released (checkpoints).
• [2026-1-28] The model implementation is released.

Inference

To run inference on audio files, use:

python inference --input_dir <input_dir> --output_dir <output_dir> [options]

Argument	Requirement / Default	Description
--input_dir	required	Path to the input directory containing audio files.
--output_dir	required	Path to the output directory where enhanced files will be saved.
--device	default: cuda:0	Torch device to run inference on, e.g., cuda:0, cuda:1, or cpu.
--extension	default: .wav	Audio file extension to process.

Training

The training script can refer to the SEtrain repository.

Citation

@ARTICLE{ulunas,
  author={Rong, Xiaobin and Yang, Leyan and Wang, Dahan and Hu, Yuxiang and Zhu, Changbao and Chen, Kai and Lu, Jing},
  journal={IEEE Transactions on Audio, Speech and Language Processing}, 
  title={UL-UNAS: Ultra-Lightweight U-Nets for Real-Time Speech Enhancement via Network Architecture Search}, 
  year={2026},
  volume={34},
  number={},
  pages={1085-1096},
  keywords={Computational modeling;Computer architecture;Convolution;Speech enhancement;Computational efficiency;Performance evaluation;Computational complexity;Artificial intelligence;Time-frequency analysis;Time-domain analysis;Speech enhancement;ultra-lightweight;neural architecture search;computational complexity},
  doi={10.1109/TASLPRO.2026.3661271}}

Contact

Xiaobin Rong: xiaobin.rong@smail.nju.edu.cn