A new cascaded spectral subtraction approach for binaural speech dereverberation and its application in source separation

Khan, Muhammad Salman; Naqvi, Syed Mohsen; Chambers, Jonathon A.

doi:10.1109/icassp.2013.6638931

Cited by 4 publications

(6 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For each sequence of the collected acoustic signal, the method in Loesch and Yang [19] is applied to estimate the number of sound sources. If the estimated number of sound sources is larger than one, which implies the acoustic signal is mixed with some interference (such as a mixture of footstep sounds and TV sounds), the SS technique proposed in Khan et al [12] is applied to separate the mixed acoustic signals to obtain the individual signal coming from each source. Moreover, the position of each sound source is also estimated, which can be taken as a cue to determine the interferences (such as the most common TV interference in a real home environment as mentioned in [41]) and can be reinforced by the fact that the prior position knowledge of an interference source (usually the TV position is fixed and can be obtained in a real home environment).…”

Section: Overview Of the Proposed Fall Detection Methods And Other Relmentioning

confidence: 99%

“…The left and right channel reverberant signals are independently processed to obtain the gain function as explained above. The two gains are combined [12] to form a single gain that is applied to the mixtures for dereverberation. The dereverberated mixtures are supplied to the second stage for separation.…”

Section: Xrev Latementioning

confidence: 99%

“…Thus, in the first stage, we dereverberate the measured two-channel acoustic mixtures. We achieve this by using a spectral subtraction based binaural dereverberation method [12].…”

Section: Dereverberationmentioning

confidence: 99%

“…reverberation time (RT60) is over 400-500 milliseconds, therefore, in the first stage we dereverberate the observed two-channel acoustic mixture. The dereverberation scheme is based on spectral subtraction [12]. In the second stage we employ an efficient model-based source separation technique which is motivated by aspects of the human auditory system by Mapped vector x in a feature space α Dual variables kðÁ; ÁÞ Kernel function combining the models of interaural level difference (ILD), interaural phase difference (IPD) and the model of mixing vectors [11].…”

Section: Interference Suppressionmentioning

confidence: 99%

See 3 more Smart Citations

An unsupervised acoustic fall detection system using source separation for sound interference suppression

et al. 2015

Self Cite

View full text Add to dashboard Cite

We present a novel unsupervised fall detection system that employs the collected acoustic signals (footstep sound signals) from an elderly person’s normal activities to construct a data description model to distinguish falls from non-falls. The measured acoustic signals are initially processed with a source separation (SS) technique to remove\ud the possible interferences from other background sound sources. Mel-frequency cepstral coeﬃcient (MFCC) features are next extracted from the processed signals and used to construct a data description model based on a one class support vector machine (OCSVM) method, which is ﬁnally applied to distinguish fall from non-fall sounds. Experiments on a recorded dataset conﬁrm that our proposed fall detection system can achieve better performance, especially with high level of interference from other sound sources, as compared with existing single microphone based methods

show abstract

Section: Overview Of the Proposed Fall Detection Methods And Other Relmentioning

confidence: 99%

Section: Xrev Latementioning

confidence: 99%

“…Thus, in the first stage, we dereverberate the measured two-channel acoustic mixtures. We achieve this by using a spectral subtraction based binaural dereverberation method [12].…”

Section: Dereverberationmentioning

confidence: 99%

Section: Interference Suppressionmentioning

confidence: 99%

See 2 more Smart Citations

An unsupervised acoustic fall detection system using source separation for sound interference suppression

et al. 2015

Self Cite

View full text Add to dashboard Cite

show abstract

“…A lot of research has been done so far in this regard. For example, the authors in [22] improved their source separation model proposed in [12], by dereverberating the audio mixtures by subtracting spectral contents of late reflections and in [23] by subtracting in cascade the spectral and the power contents of late reverberations from the speech mixtures. As our proposed model utilizes the Clifton effect, so we will put more focus here on those models which utilize this effect for deverberation of speech mixtures.…”

Section: Related Workmentioning

confidence: 99%

On enhancing model‐based expectation maximization source separation in dynamic reverberant conditions using automatic Clifton effect

Gul

Khan

Shah

et al. 2019

Int J Communication

Self Cite

View full text Add to dashboard Cite

Source separation algorithms based on spatial cues generally face two major problems. The first one is their general performance degradation in reverberant environments and the second is their inability to differentiate closely located sources due to similarity of their spatial cues. The latter problem gets amplified in highly reverberant environments as reverberations have a distorting effect on spatial cues. In this paper, we have proposed a separation algorithm, in which inside an enclosure, the distortions due to reverberations in a spatial cue based source separation algorithm namely model-based expectation-maximization source separation and localization (MESSL) are minimized by using the Precedence effect. The Precedence effect acts as a gatekeeper which restricts the reverberations entering the separation system resulting in its improved separation performance. And this effect is automatically transformed into the Clifton effect to deal with the dynamic acoustic conditions. Our proposed algorithm has shown improved performance over MESSL in all kinds of reverberant conditions including closely located sources. On average, 22.55% improvement in SDR (signal to distortion ratio) and 15% in PESQ (perceptual evaluation of speech quality) is observed by using the Clifton effect to tackle dynamic reverberant conditions.

show abstract

Enhancing the Separation of Sources in Close Proximity and Intense Reverberant Conditions by Integrating Spatial Cues with Spatial Covariance

Gul

Khan

Shah

et al. 2019

2019 15th International Conference on Emerging Technologies (ICET)

View full text Add to dashboard Cite

A new cascaded spectral subtraction approach for binaural speech dereverberation and its application in source separation

Cited by 4 publications

References 8 publications

An unsupervised acoustic fall detection system using source separation for sound interference suppression

An unsupervised acoustic fall detection system using source separation for sound interference suppression

On enhancing model‐based expectation maximization source separation in dynamic reverberant conditions using automatic Clifton effect

Enhancing the Separation of Sources in Close Proximity and Intense Reverberant Conditions by Integrating Spatial Cues with Spatial Covariance

Contact Info

Product

Resources

About