DFT-Domain Based Single-Microphone Noise Reduction for Speech Enhancement: A Survey of the State of the Art

Hendriks, Richard C.; Gerkmann, Timo; Jensen, Jesper

doi:10.2200/s00473ed1v01y201301sap011

Cited by 95 publications

(104 citation statements)

References 86 publications

Supporting

Mentioning

104

Contrasting

Order By: Relevance

“…The indices k and m denote the frequency bin and the time frame, respectively. Typically, in single-microphone approaches the speech signal X(k,m) is estimated by applying a (real-valued) gain to Y (k, m) [1]. Alternatively, in [2] [3] it has been proposed to estimate X(k,m) at each time-frequency point by applying a (complex-valued) FIR filter with the filter coefficients H l (k,m) to the noisy speech signal, i.e.,…”

Section: Multi-frame Signal Modelmentioning

confidence: 99%

See 1 more Smart Citation

Sensitivity analysis of the multi-frame MVDR filter for single-microphone speech enhancement

Fischer

Doclo

2017

2017 25th European Signal Processing Conference (EUSIPCO)

View full text Add to dashboard Cite

Abstract-Recently, a multi-frame minimum variance distortionless response (MFMVDR) filter for single-microphone noise reduction has been proposed, which exploits speech correlation across consecutive time frames. It has been shown that the MFMVDR filter achieves impressive results when the speech interframe correlation vector can be accurately estimated.In this paper, we analyze the influence of estimation errors for all required parameters, i.e., the speech interframe correlation vector and the undesired correlation matrix, on the performance of the MFMVDR filter. We compare the performance difference between oracle estimators and practically feasible blind estimators. Experimental results show that even small estimation errors substantially degrade the speech quality, where the most critical parameter is the speech interframe correlation vector.

show abstract

Section: Multi-frame Signal Modelmentioning

confidence: 99%

“…To obtain an estimate of the speech signal, typically a multiplicative real-valued gain function is applied to the noisy speech signal at each time-frequency point [1]. These approaches intrinsically assume that consecutive time frames are uncorrelated, such that each time-frequency point can be processed independently.…”

Section: Introductionmentioning

confidence: 99%

Sensitivity analysis of the multi-frame MVDR filter for single-microphone speech enhancement

Fischer

Doclo

2017

2017 25th European Signal Processing Conference (EUSIPCO)

View full text Add to dashboard Cite

show abstract

“…Directional microphones emphasize target speech that typically comes from the front by suppressing unwanted sounds coming from other spatial directions. Noise reduction systems, or more specifically, single microphone noise reduction systems, are designed to separate target speech from disturbing noise by using a separation algorithm operating on the input (Hendriks et al, 2013). This processing is most often intended to reduce gain, either in the low frequencies or in specific frequency bands, when steady-state noise (recognized as signal by the noise reduction system) is identified.…”

Section: Hearing Aids and Signal Processingmentioning

confidence: 99%

“…Thus, there is a need to develop a test which measures performance at positive or realistic SNRs (i.e., between 0 and 15 dB SNR), where traditional speech recognition threshold measures are insensitive. Moreover, hearing aid signal processing algorithms are designed to be used in positive but not negative SNRs (Hendriks et al, 2013). It is important to include a test that allows for measurement at favorable SNRs in outcome assessment.…”

Section: Measuring Hearing Aid Benefitmentioning

confidence: 99%

Cognition in Hearing Aid Users: Memory for Everyday Speech

Ng¹

2013

View full text Add to dashboard Cite

Hearing impairment interferes with speech communication. Hearing aids are designed to provide amplification for individuals with poor auditory sensitivity. Signal processing algorithms are implemented in hearing aids to further improve speech understanding and to reduce listening effort in noise. When listening effort is reduced, fewer resources are assumed to be needed for speech perception and more resources would thus be freed for cognitive processing of speech. However, this effect has only been reported in young adults with normal hearing, not in hearing aid users. Cognitive abilities, which vary between individuals, have been shown to influence the ability to benefit from hearing aids. Specifically, it is not yet known how individual differences in cognitive abilities interact with signal processing to reduce listening effort. Also, the relationship between cognition and aided speech recognition performance, as a function of acclimatization processes in new hearing aid users, has not been studied previously.This thesis investigated the importance of cognition for speech understanding in experienced and new hearing aid users. Four studies were carried out. In the first three studies (reported in Papers 1 to 3), experienced hearing aid users were tested and the aims were 1) to develop a cognitive test, called the Sentence-final Word Identification and Recall (SWIR) test, to measure the effects of a noise reduction algorithm on processing of highly intelligible speech (everyday sentences). SWIR performance is argued to reflect the amount of remaining resources upon successful speech perception; 2) to investigate, using the SWIR test, whether hearing aid signal processing would affect memory for heard speech; 3) to test whether the effects of signal processing on the ability to recall speech would interact with background noise and individual differences in working memory capacity; and 4) to explore the potential clinical application of the SWIR test by examining the relationship between SWIR performance and self-reported hearing aid outcome. In the fourth study (reported in Paper 4), the aim was 5) to examine the relationship between cognition and speech recognition in noise in new users using various models of hearing aids over the first six months of hearing aid use.Results of the studies reported in Papers 1 and 3 demonstrated that, for experienced users, noise impairs the ability to recall intelligible speech heard in noise. Noise reduction freed up cognitive resources and alleviated the negative impact of noise on memory when speech stimuli were presented in background noise consisting of speech babble spoken in the listener's native language but not in a foreign language. The possible underlying mechanisms are that noise reduction facilitates auditory stream segregation between target and irrelevant speech and reduces the attention captured by the linguistic information in irrelevant speech. In both studies, the effects of noise reduction and SWIR performance were modulated by individual differe...

show abstract

“…The availability of multiple microphones per hearing aid results in potentially more noise suppression and a better speech intelligibility [3], than with single-channel algorithms, e.g. [4]. Apart from noise suppression, it is also important to preserve the binaural cues of both desired and noise sources.…”

Section: Introductionmentioning

confidence: 99%

Improved multi-microphone noise reduction preserving binaural cues

Koutrouvelis

Hendriks

Jensen

et al. 2016

2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

We propose a new multi-microphone noise reduction technique for binaural cue preservation of the desired source and the interferers. This method is based on the linearly constrained minimum variance (LCMV) framework, where the constraints are used for the binaural cue preservation of the desired source and of multiple interferers. In this framework there is a trade-off between noise reduction and binaural cue preservation. The more constraints the LCMV uses for preserving binaural cues, the less degrees of freedom can be used for noise suppression. The recently presented binaural LCMV (BLCMV) method and the optimal BLCMV (OBLCMV) method require two constraints per interferer and introduce an additional interference rejection parameter. This unnecessarily reduces the degrees of freedom, available for noise reduction, and negatively influences the trade-off between noise reduction and binaural cue preservation. With the proposed method, binaural cue preservation is obtained using just a single constraint per interferer without the need of an interference rejection parameter. The proposed method can simultaneously achieve noise reduction and perfect binaural cue preservation of more than twice as many interferers as the BLCMV, while the OBLCMV can preserve the binaural cues of only one interferer.

show abstract

DFT-Domain Based Single-Microphone Noise Reduction for Speech Enhancement: A Survey of the State of the Art

Cited by 95 publications

References 86 publications

Sensitivity analysis of the multi-frame MVDR filter for single-microphone speech enhancement

Sensitivity analysis of the multi-frame MVDR filter for single-microphone speech enhancement

Cognition in Hearing Aid Users: Memory for Everyday Speech

Improved multi-microphone noise reduction preserving binaural cues

Contact Info

Product

Resources

About