danjacobellis/autocodec

LiVeAction: Lightweight, Versatile, and Asymmetric Codec Design for Real-time Operation

LiVeAction is a neural codec architecture designed for real-time operation on resource-constrained devices, such as wearable sensors or remote sensing platforms. It addresses the limitations of generative neural codecs by using an FFT-inspired encoder structure to reduce computational complexity and a variance-based rate penalty to enable efficient training across arbitrary signal modalities.

• Paper: LiVeAction: a Lightweight, Versatile, and Asymmetric Neural Codec Design for Real-time Operation
• Project Page: https://ut-sysml.github.io/liveaction
• Code: https://github.com/UT-SysML/liveaction
• Library: livecodec (pip install livecodec)

Citation

@inproceedings{jacobellis2026liveaction,
  title={LiVeAction: Lightweight, Versatile, and Asymmetric Codec Design for Real-time Operation},
  author={Jacobellis, Dan and Yadwadkar, Neeraja J.},
  booktitle={IEEE Data Compression Conference (DCC)},
  year={2026},
  url={https://ut-sysml.github.io/liveaction}
}