English

SpectroStream: A Versatile Neural Codec for General Audio

Sound 2025-08-08 v1 Artificial Intelligence Audio and Speech Processing

Abstract

We propose SpectroStream, a full-band multi-channel neural audio codec. Successor to the well-established SoundStream, SpectroStream extends its capability beyond 24 kHz monophonic audio and enables high-quality reconstruction of 48 kHz stereo music at bit rates of 4--16 kbps. This is accomplished with a new neural architecture that leverages audio representation in the time-frequency domain, which leads to better audio quality especially at higher sample rate. The model also uses a delayed-fusion strategy to handle multi-channel audio, which is crucial in balancing per-channel acoustic quality and cross-channel phase consistency.

Cite

@article{arxiv.2508.05207,
  title  = {SpectroStream: A Versatile Neural Codec for General Audio},
  author = {Yunpeng Li and Kehang Han and Brian McWilliams and Zalan Borsos and Marco Tagliasacchi},
  journal= {arXiv preprint arXiv:2508.05207},
  year   = {2025}
}
R2 v1 2026-07-01T04:38:45.523Z