Beschreibung: 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 approach.

Beschreibung: A priori signal-to-noise ratio (SNR) contains critical information about the single-channel mixture of a speech and noise signal, and can be used by speech enhancement algorithms. We propose a novel a priori SNR estimator using the estimates obtained from discriminative non-negative matrix factorization (DNMF). The idea of our new approach is to utilize the DNMF to perform the preliminary speech components estimation, which can be either directly used to estimate the a priori SNR, or can be combined with the well-known decision-directed (DD) approach by Ephraim and Malah to perform the a priori SNR estimation. We present a speaker-independent but noise-dependent DNMF-based a priori SNR estimator. Speech enhancement simulation results in the presence of non-stationary noise validate our new approach combined with well-known spectral weighting rules, outperform several NMF-based and non-NMF-based state-of-the-art methods. Our new a priori SNR estimation approach also offers more balanced behavior compared to all baseline methods and is more independent of the weighting rules by giving a more precise a priori SNR estimation. In the low-SNR condition (SNR=-5dB), our approach shows a 1 to 2 dB higher SNR improvement with a similar or better PESQ MOS than all the baseline methods.