Concatenated Identical DNN (CI-DNN) to Reduce Noise-Type Dependence in DNN-Based Speech Enhancement

TL;DR

This paper introduces a concatenated identical DNN framework for speech enhancement that improves noise robustness and speech quality, outperforming traditional methods and generalizing well to unseen noise types.

Contribution

The paper proposes a novel CI-DNN architecture trained under multiple SNR conditions, enhancing noise reduction and speech quality with fewer parameters and better generalization.

Findings

Outperforms classical spectral weighting in speech quality and intelligibility

Achieves similar or better performance with fewer trainable parameters

Generalizes better to unseen noise types compared to other deep learning approaches

Abstract

Estimating time-frequency domain masks for speech enhancement using deep learning approaches has recently become a popular field of research. In this paper, we propose a mask-based speech enhancement framework by using concatenated identical deep neural networks (CI-DNNs). The idea is that a single DNN is trained under multiple input and output signal-to-noise power ratio (SNR) conditions, using targets that provide a moderate SNR gain with respect to the input and therefore achieve a balance between speech component quality and noise suppression. We concatenate this single DNN several times without any retraining to provide enough noise attenuation. Simulation results show that our proposed CI-DNN outperforms enhancement methods using classical spectral weighting rules w.r.t. total speech quality and speech intelligibility. Moreover, our approach shows similar or even a little bit better performance with much fewer trainable parameters compared with a noisy-target single DNN approach of the same size. A comparison to the conventional clean-target single DNN approach shows that our proposed CI-DNN is better in speech component quality and much better in residual noise component quality. Most importantly, our new CI-DNN generalized best to an unseen noise type, if compared to the other tested deep learning approaches.