A New Speech Enhancement Technique to Reduce Residual Noise Using Perceptual Constrained Spectral Weighted Factors

Transcription

1 IOSR Journal of Electronics an Communication Engineering (IOSR-JECE) e-issn: ,p- ISSN: Volume 6, Issue 3 (May. - Jun. 013), PP 8-33 A New Speech Enhancement Technique to Reuce Resiual Noise Using Perceptual Constraine Spectral Weighte Factors K.Ravi Kumar, T.Munikumar Faculty Member, Dept.of ECE,Gulavalleru Engineering College, Gulavalleru-51356, AP, Inia Faculty Member, Dept.of ECE, Amrita sai Engineering College, Paritala-51356, AP, Inia. Abstract- This paper eals with resiual musical noise which results from the perceptual speech enhancement type algorithms an especially using wiener filtering approach. Perceptual speech enhancement techniques perform better than the non perceptual techniques, most of them still return a trouble resiual musical noise. This is ue to that only noise above the noise masking threshol (NMT) is filtere out then noise below the noise masking threshol (NMT) can become auible if its maskers are filtere. It can affect the performance of perceptual speech enhancement metho that process the auible noise only (Resiual noise is still present). In orer to overcome this rawback a new speech enhancement technique is propose here.the main aim here is to improve the enhance speech signal quality provie by perceptual wiener filtering an by controlling the latter via a secon filter regare as a psychoacoustically motivate weighting factor. The simulation results gives the information that the performance is improve compare to other perceptual speech enhancement methos. I. Introuction The objective of speech enhancement is to improve the intelligibility an quality of speech in noisy environments. Many approaches have been propose like spectral subtractive type [1-4], Perceptual Wiener filtering types. Among them spectral subtraction an the Wiener filtering algorithms are using frequently because of their low computational complexity an outstaning performance. In these algorithms, Such methos leaves resiual noise known as musical noise. This type of noise is quite annoying(trouble). In orer to reuce the effect of musical noise, several methos an solutions have been propose. Some of them involve ajusting the parameters of spectral subtraction so as to give more flexibility as in [] an [3]. Other techniques such as propose in [4], are base on signal subspace approaches. Even though these methos effectively improve the signal to noise ratio (SNR), the problem of eliminating musical noise is still a challenge to many researchers. In the last few ecaes the introuction of moeling the signal with consieration of psychoacoustic has attracte a great eal of interest. The main objective is to improve the perceptual quality of the enhance signal. In [3], a psychoacoustic moel is use to change or control the parameters of the spectral subtraction in orer to fin the best trae off between speech istortion an noise reuction. To make musical noise inauible, the propose technique linear estimator in [5] uses the human auitory system an its masking properties. In [6], the intermeiate signal an masking threshol, which is slightly enoise an free of musical noise, are use to etect musical tones generate by the spectral subtraction methos. This etection can be use by a postprocessing techniques which aims at reucing the etecte tones. These perceptual speech enhancement systems reuce the resiual noise but introuce some unesire istortion to the enhance speech signal. When this istorte estimate speech signal(enhance) is applie to the recognition systems their performance egraes rastically. The iea of the propose metho is to remove, perceptually significant noise components from the noisy or corrupte signal, so that the clean speech components are not affecte by processing (any enhancing process). In aition, the technique requires very little a priori information (information before processing the noisy) signal of the features of the noise. In the present paper, we propose to control the perceptual wiener filtering by psychoacoustically motivate filter that can be regare as weighting factor. The purpose is to minimize or reuce the perception of resiual noise without egraing the clarity of the enhance speech signal. II. stanar speech enhancement technique Let the noisy signal can be expresse as y( x( (, (1) Where x ( is the original clean speech signal an ( is the aitive ranom noise signal, uncorrelate with the original signal. Taking DFT to the observe signal gives Y( X ( D(. () 8 Page

2 A New Speech Enhancement Technique To Reuce Resiual Noise Using Perceptual Constraine Where m 1,,..., M is the frame inex, k 1,,..., K is the frequency bin inex, M is the total number of frames an K is the frame length, Y(, X ( an D( represent the short time spectral components of the y (, x( an(, respectively. Clean speech spectrum Xˆ ( is obtaine by multiplying noisy speech spectrum with filter gain function as given in eqation (3) Xˆ ( H( Y( (3) Where H ( is the noise suppression filter gain function Wiener filter (WF)), which is erive accoring to MMSE estimator an H( is given by ( H( (4) 1 ( Where ( is an apriori SNR, which is efine as x ( (. (5) ( E D( an x( E x( represents the ( estimate noise power spectrum an clean speech power spectrum. A posteriori estimation is given by Y( ( (6) ( An estimate of ˆ( as of ( H( m 1, Y( m 1, ˆ( (1 ) P' V ( k. (7) is given by the well known ecision irecte approach [9] an is expresse Where V( ( 1, Px x if x 0 an P x 0 otherwise. The noise suppression gain function is chosen as the Wiener filter same as in[13] III. Perceptual Speech Enhancement Although the Wiener filtering reuces the level of musical noise, it oes not eliminate it [15]. Musical noise exists an perceptually troubles. So as an effort to make the resiual noise perceptually inauible, many perceptual speech enhancement methos have been propose which incorporates the auitory masking properties [-9]. In these methos resiual noise is shape accoring to an estimate of the signal masking threshol [9, 13]. Figure 1 epicts the complete block iagram of the propose speech enhancement metho. Figure1. Block iagram of the propose speech enhancement metho 9 Page

3 A New Speech Enhancement Technique To Reuce Resiual Noise Using Perceptual Constraine 3.1 Gain of Perceptual Wiener filter (PWF) The perceptual Wiener filter (PWF) gain function H 1( is calculate base on cost function, J which is efine as J Xˆ ( X ( (8) Substituting () an (3) in (9) results to Where i E ( H k 1 ( 1) X ( H1( D( ) i r i (9) ( H1 ( 1) EX ( an H1 ( ED( r i represents speech istortion energy an resiual noise energy. To make this resiual noise inauible, the resiual noise shoul be less than the auitory masking threshol,. This constraint is given by T ( r ( T (10) i By incluing the above constraint an substituting ( ED( ( E X ( in (9) the cost function will become as x J ( H1( 1) s ( H1 ( max ( T(,0 (11) The esire perceptual moification of Wiener is obtaine by ifferentiating J w.r.t H1( an equating to zero. The obtaine perceptually efine Wiener filter gain function is given by ( x ( H1 (1) ( max( ( T(,0) x By multiplying an iviing equation (1) with (, H ( ) will become as ˆ( H1( max( ( T(,0) ˆ( ( T( (13) 1 k is noise masking threshol which is estimate base on[16] noisy speech spectrum. A priori SNR an noise power spectrum were estimate using the two -step a priori SNR estimator propose in [15] an weighte noise estimation metho propose in[17],respectively. 3. WEIGHTED PWF Although perceptual speech enhancement methos perform better than the non-perceptual methos, most of them still return annoying resiual musical noise. Enhance speech signal obtaine using above mentione perceptual Wiener filter still contains some resiual noise ue to the fact that only noise above the noise masking threshol is filtere an noise below the noise masking threshol is remain. It can affect the performance of perceptual speech enhancement metho that processes auible noise only. In orer to overcome this rawback we propose to weight the perceptual Wiener filters using a psychoacoustically motivate weighting filter. Psychoacoustically motivate weighting filter is given by H(, ifath ( T( W ( (15) 1, otherwise an Where ATH ( is the absolute threshol of hearing. This weighting factor is use to weight the perceptual wiener filter. The gain function of the H ( ) of the propose weighte perceptual Wiener filter is given by k H H1( W( (16) 30 Page

4 A New Speech Enhancement Technique To Reuce Resiual Noise Using Perceptual Constraine IV. Simulation Results To evaluate the performance of the propose scheme of speech enhancement an for comparision, simulations are carrie out with the NOIZEUS, A noisy speech corpus for evaluation of speech enhancement algorithms, atabase [18]. The noisy atabase contains 30 IEEE sentences (prouce by three female an three male speakers) corrupte by eight ifferent real worl noises at ifferent SNRs levels. Speech signals were egrae with ifferent types of noise at global SNR levels of 0 B, 5 B, 10 B an 15 B. In this evaluation only five noises are consiere those are car, babble, airport, train, an street noise. The objective quality measures use for the evaluation of the propose speech enhancement metho are the PESQ measures an segmental SNR [19]. It is well known that the segmental SNR is more accurate in inicating the speech istortion than the overall SNR. The higher value of the segmental SNR inicates the weaker speech istortion ie less istortion. The higher PESQ score inicates better perceive quality of the propose signal [19]. The performance of the propose metho is compare with Wiener filter an perceptual Wiener filter. The simulation results are summarize in Table 1 an Table. The propose metho leas to better improvements are obtaine for the high noise level an the better enoising quality for temporal. The timefrequency istribution of speech signals provies more accurate information about the resiual musical noise an speech istortion than the corresponing time omain waveforms. Here we compare the spectrograms for each of the techniques an confirme a reuction of the speech istortion an resiual noise. Figure. Represents the spectrograms of the noisy signal, clean speech signal, an enhance speech signals. Table.1 Segmental SNR values of Enhance Signals Noise Type Babble Car Train Airport Street Input SNR (B) WF PWF Weighte PWF Table. PESQ values of the enhance signals Noise Type Input WF PWF Weighte PWF SNR (B) Babble Car Train Airport Street Page

5 A New Speech Enhancement Technique To Reuce Resiual Noise Using Perceptual Constraine V. Conclusion In this paper, an effective approach for suppressing the musical noise presente after wiener filtering has been introuce. Base on the perceptual properties of the human auitory syste a weighting factor accentuates the enoising process when noise is perceptually insignificant an prevents that resiual noise components might become auible in the absence of ajacent maskers. When the speech signal is aitively corrupte by babble noise an car noise objective measure results showe the improvement brought by the propose metho in comparison to some recent filtering techniques of the same type. Figure. speech spectrogra(a)original clean signal,(b) noisy signal(babble noise SNR=5B),(c)enhance signal using Wiener filter()enhance signal using PWF,(e)enhance signal using Weighte PWF References [1] Y. Ephraim an D. Malah, Speech enhancement using a minimum mean square error short-time spectral amplitue estimator, IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-3, pp , Dec [] R. Schwartz M. Berouti an J. Makhoul, Enhancement of speech corrupte by acoustic noise, Proc. of ICASSP, 1979, vol. I, pp [3] N.Virag, Single channel speech enhancement base on masking properties of the human auitory syste IEEE Trans. Speech an Auio Processing, vol. 7, pp , [4] Y. Ephraim an H.L. Van Trees, A signal subspace approach for speech enhancement, IEEE Trans. Speech an Auio Processing, vol. 3, pp , [5] Y. Hu an P. Loizou, Incorporating a psychoacoustic moel in frequency omain speech enhancement, IEEE Signal Processing Letters, vol. 11(), pp , 004. [6] F. Jabloun an B. Champagne, Incorporating the human hearing properties in the signal subspace approach for speech enhancement, IEEE Trans. Speech an Auio Processing,vol. 11, pp , 003. [7] Y.M. Cheng an D. O Shaughnessy, Speech enhancement base conceptually on auitory evience, IEEE Trans. Signal Processing, vol.39, no.9, pp , [8] D. Tsoukalas, M. Paraskevas, an J. Mourjopoulos, Speech enhancement using psychoacoustic criteria, IEEE ICASSP, pp , Minneapolis, MN, Page

6 A New Speech Enhancement Technique To Reuce Resiual Noise Using Perceptual Constraine [9] Y. Hu an P.C. Loizou, "A perceptually motivate approach for speech enhancement," IEEE Trans. Speech Auio Processing, pp Sept [10] L. Lin, W. H. Holmes an E. Ambikairajah, Speech enoising using perceptual moification of Wiener filtering, IEE Electronic Letters, vol. 38, pp , Nov 00. [11] P. Scalart C. Beaugeant, V. Turbin an A. Gilloire, New optimal filtering approaches for hans-free telecommunication terminals, Signal Processing, vol. 64 (15), pp , Jan [1] T. Lee an Kaisheng Yao, Speech enhancement by perceptual filter with sequential noise parameter estimation, Proc. of ICASSP, vol. I, pp , 004. [13] M. Jahangir Ala Si-Ahme Selouani, Douglas O Shaughnessy an S. Ben Jebara, Speech enhancement using a Wiener enoising technique an musical noise reuction in the Proceeing of INTERSPEECH 08, Brisbane, Australia, pp , September 008. [14] Amehraye, D. Pastor, an A. Tamtaoui, Perceptual improvement of Wiener filtering. Proc. of ICASSP, pp , 008. [15] M. Jahangir Ala Douglas O Shaughnessy an Si-Ahme Selouani, Speech enhancement base on novel two-step a priori SN estimators, in the Proceeing of INTERSPEECH 08, Brisbane,Australia, pp , September 008. [16] J. D. Johnston, Transform coing of auio signals using perceptual noise criteria, IEEE on Selecte Areas in Comm., vol. 6, pp , February1988. [17] M. Kato, A. Sugiyama an M. Serizawa, Noise suppression with high speech quality base on weighte noise estimation an MMSESTSA, IEICE Trans. Funamentals, vol. E85-A, no.7, pp , July 00. [18] [19] Yi Hu an Philipos C. Loizou, Evaluation of Objective Quality Measures for Speech Enhancement, IEEE Trans. on Auio, Speech an Language Processing, vol. 16, no. 1, pp. 9-38, January Page