SACodec: Asymmetric Quantization with Semantic Anchoring for Low-Bitrate High-Fidelity Neural Speech Codecs
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (36) · 30539-30547
DOI: 10.1609/aaai.v40i36.40308 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Zhongren Dong und 6 weiteren Autor:innen.
Abstract (Originalsprache)
Neural Speech Codecs face a fundamental trade-off at low bitrates: preserving acoustic fidelity often compromises semantic richness. To address this, we introduce SACodec, a novel codec built upon an asymmetric dual-quantizer that employs our proposed Semantic Anchoring mechanism. This design strategically decouples the quantization of Semantic and Acoustic details. The semantic anchoring is achieved via a lightweight projector that aligns acoustic features with a frozen, large-scale mHuBERT codebook, injecting linguistic priors while guaranteeing full codebook utilization. Sequentially, for acoustic details, a residual activation module with SimVQ enables a single-layer quantizer (acoustic path) to faithfully recover fine-grained information. At just 1.5 kbps, SACodec establishes a new state of the art by excelling in both fidelity and semantics: subjective listening tests confirm that its reconstruction quality is perceptually highly comparable to ground-truth audio, while its tokens demonstrate substantially improved semantic richness in downstream tasks. This work suggests that assigning specialized semantic quantizers to distinct information streams offers an effective path to reconcile the long-standing trade-off between fidelity, semantics, and modeling simplicity in low-bitrate speech tokenization.
Zitieren
APA
Dong, Z., Wang, B., Han, J., Guo, H., Mo, X., Cao, Y., & Zhang, Z. (2026). SACodec: Asymmetric Quantization with Semantic Anchoring for Low-Bitrate High-Fidelity Neural Speech Codecs. Proceedings of the AAAI Conference on Artificial Intelligence, 40(36), 30539-30547. https://doi.org/10.1609/aaai.v40i36.40308
Vancouver
Dong Z, Wang B, Han J, Guo H, Mo X, Cao Y, et al. SACodec: Asymmetric Quantization with Semantic Anchoring for Low-Bitrate High-Fidelity Neural Speech Codecs. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(36):30539-30547. doi:10.1609/aaai.v40i36.40308
BibTeX
@article{Dong2026,
author = {Dong, Zhongren and Wang, Bin and Han, Jing and Guo, Haotian and Mo, Xiaojun and Cao, Yimin and Zhang, Zixing},
title = {SACodec: Asymmetric Quantization with Semantic Anchoring for Low-Bitrate High-Fidelity Neural Speech Codecs},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {36},
pages = {30539--30547},
doi = {10.1609/aaai.v40i36.40308}
}
EUVIMED