EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Robust speech denoising using a parallel target-field non-causal WaveNet under stationary and non-stationary noise

Jella Sandhya, Shafiq Ur Rehman, Shaik Mazhar Hussain

Sound & Vibration · 23.07.2026

DOI: 10.59400/sv4065 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sound & Vibration, online erschienen am 23.07.2026. Verfasst von Jella Sandhya und 2 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The presence of stationary or non-stationary noise substantially impairs speech intelligibility and its perceptual quality in present-day communication systems. Many existing spectral-domain methods like Wiener filtering and spectral subtraction rely on strong statistical assumptions about complex-valued data, while state-of-the-art neural models such as Conv-TasNet, DCCRN, and MetricGAN+ achieve remarkable performance gains for speech enhancement, but tend to suffer from heavy computational complexity and inference latency due to their complex architecture. In this paper, we present a new supervised non-causal WaveNet with parallel target-field prediction for efficient end-to-end speech denoising. The proposed model captures both past and future temporal context by enabling symmetric receptive fields and removing autoregressive dependencies. Traditional autoregressive WaveNet models generate samples one at a time, which require considerable redundant convolution operations due to the repetitive nature of their structures and tasks, while the presented method is capable of predicting a target field of samples in a single forward pass, allowing for parallel inference. We validate the proposed model on the NSDTSEA dataset for input SNR levels of 2.5 dB to 17.5 dB and under stationary as well as non-stationary noise types. We showcase experimental results that prove that the proposed framework consistently outperforms strong classical baselines, with over 1.22 dB gain in output SNR and 46% reduction in Mel Cepstral Distortion (MCD) under non-stationary noise. Moreover, when compared to state-of-the-art deep learning models such as Conv-TasNet, DCCRN, and MetricGAN+, the proposed method achieves a viable trade-off between performance, robustness, and computational efficiency.

Zitieren

APA

Sandhya, J., Rehman, S. U., & Hussain, S. M. (2026). Robust speech denoising using a parallel target-field non-causal WaveNet under stationary and non-stationary noise. Sound & Vibration. https://doi.org/10.59400/sv4065

Vancouver

Sandhya J, Rehman SU, Hussain SM. Robust speech denoising using a parallel target-field non-causal WaveNet under stationary and non-stationary noise. Sound & Vibration. 2026. doi:10.59400/sv4065

BibTeX
@article{Sandhya2026,
  author = {Sandhya, Jella and Rehman, Shafiq Ur and Hussain, Shaik Mazhar},
  title = {Robust speech denoising using a parallel target-field non-causal WaveNet under stationary and non-stationary noise},
  journal = {Sound & Vibration},
  year = {2026},
  doi = {10.59400/sv4065}
}
Kostenlos starten →