A novel fast nonstationary noise tracking approach based on MMSE spectral power estimator

Zhang, Qiquan; Wang, Mingjiang; Lu, Yun; Zhang, Lu; Idrees, Muhammad

doi:10.1016/j.dsp.2019.01.019

Cited by 18 publications

(10 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The MMSE-based noise PSD estimators reported in [19]- [23] are briefly summarised in this section. They are all based on an MMSE estimate of the noise periodogram, which is defined as the conditional expectation: E{N 2 |Y, λ s , λ d }.…”

Section: Overview Of Mmse-based Noise Psd Estimatorsmentioning

confidence: 99%

“…From Equation ( 9), it is obvious that the MMSE-based Bayesian noise PSD estimator depends on the a priori and a posteriori SNR. As shown in [19]- [23], the noise PSD tracking performance is predominately affected by the accuracy of the a priori SNR estimate,ξ.…”

Section: Overview Of Mmse-based Noise Psd Estimatorsmentioning

confidence: 99%

“…The SPP was calculated using a fixed optimal a priori SNR. In [23], an MMSE speech spectral power estimator that incorporates speech presence uncertainty (SPU) and a bias correction factor was proposed to improve the DD approach, producing an improved tracking capability (ImMMSE).…”

Section: Overview Of Mmse-based Noise Psd Estimatorsmentioning

confidence: 99%

“…2) Next, the a posteriori SNR estimate,γ = Y 2 / λ d , is used to find the a priori SNR estimate. As in [14], [17], [20], [22], [23], the DD approach is used to estimate the a priori SNR for the MS, MCRA-2, MMSE, SPP, and ImMMSE methods. * * The smallest allowable value is set to ξ min = −15 dB for the DD approach, as in [26].…”

Section: E Evaluation Measuresmentioning

confidence: 99%

“…Another class of noise PSD trackers is formulated from the perspective of Bayesian statistics. Several minimum meansquare error (MMSE)-based Bayesian noise PSD estimators exist in the literature [19]- [23], and are described in Section III. These noise trackers are based on an MMSE estimate of the noise periodogram, which is computed using the a priori and a posteriori SNR.…”

mentioning

confidence: 99%

See 4 more Smart Citations

DeepMMSE: A Deep Learning Approach to MMSE-Based Noise Power Spectral Density Estimation

Zhang

Nicolson

Wang

et al. 2020

IEEE/ACM Trans. Audio Speech Lang. Process.

Self Cite

110

View full text Add to dashboard Cite

An accurate noise power spectral density (PSD) tracker is an indispensable component of a single-channel speech enhancement system. Bayesian-motivated minimum mean-square error (MMSE)-based noise PSD estimators have been the most prominent in recent time. However, they lack the ability to track highly non-stationary noise sources due to current methods of a priori signal-to-noise (SNR) estimation. This is caused by the underlying assumption that the noise signal changes at a slower rate than the speech signal. As a result, MMSE-based noise PSD trackers exhibit a large tracking delay and produce noise PSD estimates that require bias compensation. Motivated by this, we propose an MMSE-based noise PSD tracker that employs a temporal convolutional network (TCN) a priori SNR estimator. The proposed noise PSD tracker, called DeepMMSE makes no assumptions about the characteristics of the noise or the speech, exhibits no tracking delay, and produces an accurate estimate that requires no bias correction. Our extensive experimental investigation shows that the proposed DeepMMSE method outperforms state-of-the-art noise PSD trackers and demonstrates the ability to track abrupt changes in the noise level. Furthermore, when employed in a speech enhancement framework, the proposed DeepMMSE method is able to outperform state-of-the-art noise PSD trackers, as well as multiple deep learning approaches to speech enhancement.

show abstract

Section: Overview Of Mmse-based Noise Psd Estimatorsmentioning

confidence: 99%

Section: Overview Of Mmse-based Noise Psd Estimatorsmentioning

confidence: 99%

Section: Overview Of Mmse-based Noise Psd Estimatorsmentioning

confidence: 99%