EUVIMEDEUVIMED entdecken
Crossref

Enhancing Cough Sound Detection by Wavelet Scattering Transform

Yiliang ZHAO, Zhong JI, Huaxing XU, Haichuan REN

IEICE Transactions on Information and Systems · 2026

DOI: 10.1587/transinf.2026edp7113 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEICE Transactions on Information and Systems, 2026. Verfasst von Yiliang ZHAO und 3 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

A robust and effective acoustic feature is crucial for audio-based cough detection in complex environments. As a highly effective feature extraction method, wavelet scattering transform (WST) generates representations that enjoy stability properties with respect to small translation and deformations, making it highly resilient to noise and variability. This paper integrates the wavelet scattering transform into the audio spectrogram Transformer model for cough detection and evaluates its performance for the first time. We evaluate the abilities of WST on a publicly available realistic cough dataset and show that WST slightly outperforms the Melspectrogram on Accuracy and AUC, and clearly outperforms MFCC. These results underscore the effectiveness of WST in handling challenging acoustic cough environments, establishing it as a promising candidate feature extractor for analyzing cough signals.

Zitieren

APA

ZHAO, Y., JI, Z., XU, H., & REN, H. (2026). Enhancing Cough Sound Detection by Wavelet Scattering Transform. IEICE Transactions on Information and Systems. https://doi.org/10.1587/transinf.2026edp7113

Vancouver

ZHAO Y, JI Z, XU H, REN H. Enhancing Cough Sound Detection by Wavelet Scattering Transform. IEICE Transactions on Information and Systems. 2026. doi:10.1587/transinf.2026edp7113

BibTeX
@article{ZHAO2026,
  author = {ZHAO, Yiliang and JI, Zhong and XU, Huaxing and REN, Haichuan},
  title = {Enhancing Cough Sound Detection by Wavelet Scattering Transform},
  journal = {IEICE Transactions on Information and Systems},
  year = {2026},
  doi = {10.1587/transinf.2026edp7113}
}
Kostenlos starten →