EUVIMEDEUVIMED entdecken
Crossref

DegVoC: Revisiting Neural Vocoder from a Degradation Perspective

Andong Li, Tong Lei, Lingling Dai, Kai Li, Rilin Chen, Meng Yu, Xiaodong Li, Dong Yu, Chengshi Zheng

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (37) · 31510-31518

DOI: 10.1609/aaai.v40i37.40416 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Andong Li und 8 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Existing neural vocoders have demonstrated promising performance by leveraging Mel-spectrum as an acoustic feature for conditional audio generation. Nonetheless, they remain constrained by an inherent ``performance-cost'' dilemma that significantly hinders the development of this field. This paper revisits this foundational task from a novel degradation perspective, where Mel-spectrum is regarded as a special signal degradation process from the target spectrum. Drawing inspiration from traditional sparse signal recovery problems, we propose DegVoC, a GAN-based neural vocoder with a two-step solution procedure. First, by exploiting degradation priors, we attempt to retrieve the initial spectral structure from Mel-domain representations as an initial solution via a simple linear transformation. Based on that, we introduce a deep prior solver that accounts for the heterogeneous distribution of sub-bands in the time-frequency domain. A convolution-style attention module with a large kernel size is specially devised for efficient inter-frame and inter-band contextual modeling. With 3.89 M parameters and substantially reduced inference complexity, DegVoC achieves state-of-the-art performance across objective and subjective evaluations, outperforming existing GAN-, DDPM- and flow-matching-based baselines.

Zitieren

APA

Li, A., Lei, T., Dai, L., Li, K., Chen, R., Yu, M., Li, X., Yu, D., & Zheng, C. (2026). DegVoC: Revisiting Neural Vocoder from a Degradation Perspective. Proceedings of the AAAI Conference on Artificial Intelligence, 40(37), 31510-31518. https://doi.org/10.1609/aaai.v40i37.40416

Vancouver

Li A, Lei T, Dai L, Li K, Chen R, Yu M, et al. DegVoC: Revisiting Neural Vocoder from a Degradation Perspective. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(37):31510-31518. doi:10.1609/aaai.v40i37.40416

BibTeX
@article{Li2026,
  author = {Li, Andong and Lei, Tong and Dai, Lingling and Li, Kai and Chen, Rilin and Yu, Meng and Li, Xiaodong and Yu, Dong and Zheng, Chengshi},
  title = {DegVoC: Revisiting Neural Vocoder from a Degradation Perspective},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {37},
  pages = {31510--31518},
  doi = {10.1609/aaai.v40i37.40416}
}
Kostenlos starten →