Split Bregman Approach to Linear Prediction Based Dereverberation With Enforced Speech Sparsity

Witkowski, Marcin; Kowalczyk, Konrad

doi:10.1109/lsp.2021.3070734

Cited by 10 publications

(2 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One such method is the weighted prediction error (WPE) method [11], which has demonstrated its effectiveness in dereverberation. To further enhance performance, works such as [16] and [17] propose leveraging speech sparsity in the time-frequency domain to incorporate an additional prior on the unknown variance. Despite the clear physical interpretation of these model-based methods, their performance and robustness are limited as they do not fully exploit the inherent priors of speech signal structures.…”

Section: Introductionmentioning

confidence: 99%

“…Specifically, we formulate the prediction error minimization problem of WPE with an additional regularizer that is not explicitly handcrafted. Unlike vanilla WPE and its extensions [16], [17], [34], [35], which do not consider sophisticated speech priors, we propose integrating speech prior information by employing the plugand-play strategy. Specifically, we employ the regularization by denoising (RED) strategy, which is a promising variant of PnP [36].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Virtual Acoustic Channel Expansion Based on Neural Networks for Weighted Prediction Error-Based Speech Dereverberation

Yang¹,

Chang²

2020

Interspeech 2020

View full text Add to dashboard Cite

Speech dereverberation is an important issue for many real-world speech processing applications. Among the techniques developed, the weighted prediction error (WPE) algorithm has been widely adopted and advanced over the last decade, which blindly cancels out the late reverberation component from the reverberant mixture of microphone signals. In this study, we extend the neural-network-based virtual acoustic channel expansion (VACE) framework for the WPE-based speech dereverberation, a variant of the WPE that we recently proposed to enable the use of dual-channel WPE algorithm in a single-microphone speech dereverberation scenario. Based on the previous study, some ablation studies are conducted regarding the constituents of the VACE-WPE in an offline processing scenario. These studies help understand the dynamics of the system, thereby simplifying the architecture and leading to the introduction of new strategies for training the neural network for the VACE. Experimental results in noisy reverberant environments reveal that VACE-WPE considerably outperforms its single-channel counterpart in terms of objective speech quality and is complementary to the singlechannel WPE when employed as the front-end for the far-field automatic speech recognizer.

show abstract

Section: Introductionmentioning

confidence: 99%