Optimal Filter Bank Research Articles

Filterbank Optimization for Text-Dependent Speaker Verification by Evolutionary Algorithm Using Spline-Defined Design Parameters

Mel frequency cepstral coefficients (MFCCs) have been the most predominantly used spectral features in many a speech-based application. It was primarily introduced to address speech recognition and was later adopted for various other applications such as speaker recognition and emotion recognition. Several findings, in recent times, suggest that Mel-scale filterbank, which is primarily inspired by human perception phenomenon, may not be the most optimum one for speaker recognition. Working in the same direction, this study attempts optimization of filterbank design for text-dependent speaker verification. Motivated by the success of evolutionary computations in the related fields, an evolutionary algorithm is used to carry out this optimization process. This brings into effect data-driven learning of the design parameters and is hypothesized to yield filterbanks which would suit the specific task of speaker-phrase discrimination. The filterbanks have been optimized for the task of text-dependent speaker verification in general, and also for specific cases of speakers and phrases. The proposed filterbank results in relative equal error rate reduction of up to 39.41% with respect to the baseline MFCCs.

Fast-convolution multicarrier based frequency division multiple access

Fast-convolution multicarrier (FCMC), the asynchronous waveform with ultra-low sidelobe, has appeared to be a promising waveform technique for future wireless communications. In this paper, we investigate the filter bank optimization as well as receiver design including low-complexity channel estimator and equalizer for FCMC based frequency division multiple access (FDMA). Starting from the conventional multi-carrier signals, we first derive a vectorized signal model as a framework to systematically design the FCMC transceiver. For nearly perfect reconstruction (NPR) filter banks design, an optimization criterion which consists of minimizing received signal segments mean square error (MSE) is proposed. From the fact that Toeplitz matrices can be asymptotically diagonalized by discrete Fourier transform (DFT) matrix, the channel equalizer can be simplified to one-tap frequency domain equalizer when DFT size is large enough. The minimum mean square error (MMSE) criterion is then applied to calculate the coefficients of one-tap frequency domain equalizer. In practice, due to the channel fading, the channel estimation has to be performed to obtain the channel state information (CSI) which is required by the channel equalization. To this end, we propose a combined cyclic prefix (CP) and cyclic suffix (CS) pilot structure which facilitates to estimate the frequency domain CSI directly in the receiver end. The proposed FCMC based FDMA features low-complexity receiver, adjustable users bandwidth and low peak-to-average power ratio. Simulation results confirm the performance of the proposed scheme.

Detection of shockable ventricular arrhythmia using optimal orthogonal wavelet filters

Sudden cardiac death (SCD) is caused by lethal arrhythmia. Ventricular fibrillation (VF) and ventricular tachycardia (VT) are amenable to defibrillation or electrical shock therapy (“shockable” arrhythmia) that can abolish the VF/VT and restore normal electrical and mechanical heart function. The challenge is to differentiate between shockable and non-shockable arrhythmia during the emergency response to SCD. When it comes to saving the life, accurate electrocardiogram (ECG) diagnosis and fast delivery of appropriate treatment is imperative. Automated systems to differentiate shockable from non-shockable arrhythmia have been developed to overcome the difficulty, and possible errors due to the manual inspection. In the present work, we have devised an efficient, effective and robust automated system to detect shockable and non-shockable arrhythmia using an optimal wavelet-based features extracted from ECG epochs of 2 s durations. We employed optimal two-channel frequency selective orthogonal wavelet filter bank to diagnose shockable ventricular arrhythmia. The optimization was carried out by minimizing the stop band ripple energy of the wavelet filter. The optimal orthogonal wavelet filter has been designed using a semi-definite programming (SDP) formulation without the use of any parameterization. The SDP solution gave us the desired optimal orthogonal wavelet filter bank with minimum stop band energy and the desired degree of regularity for the given length of filter. Fuzzy entropy and Renyi entropy features were extracted from the 2-s ECG epochs. These extracted features were then fed into the classifiers for discrimination of shockable arrhythmia rhythms and non-shockable arrhythmia rhythms. The best results were obtained from support vector machine. Accuracy of 97.8%, sensitivity of 93.42%, and specificity of 98.35% were obtained using a tenfold cross validation scheme. The developed automated system is accurate and robust; therefore, it can be integrated in automated external defibrillators that can be deployed for hospitals as well as out-of-hospital emergency resuscitation of SCD.

Decimations of intrinsic mode functions via semi-infinite programming based optimal adaptive nonuniform filter bank design approach

Abstract A signal can be represented as the sum of the intrinsic mode functions via performing the empirical mode decomposition. For the discrete time signals, the lengths of the intrinsic mode functions are equal to the lengths of the input signals. As there is usually more than one intrinsic mode function, the total numbers of discrete points of all the intrinsic mode functions are usually more than the lengths of the input signals. In other words, the empirical mode decomposition is an oversampled representation. For some applications such as the compression application, the oversampled representation is not preferred. Therefore, this paper proposes an optimal adaptive nonuniform filter bank design approach for performing the decimations on the intrinsic mode functions. In particular, the passbands of the intrinsic mode functions are estimated based on an adaptive gradient algorithm. Then, the intrinsic mode functions are nonuniformly downsampled and upsampled with the sampling integers derived based on their estimated passbands. Next, a bank of filters is employed to reconstruct the original signal. Here, the passbands of the filters are also derived based on those of the intrinsic mode functions. After that, the nonuniform filter bank design problem is formulated as a semi-infinite programming problem such that the total ripple energies of all the filters are minimized subject to the specifications on the absolute maximum values of both the real part and the imaginary part of the reconstruction error. The semi-infinite programming problem is approximated by a semi-definite programming problem. Computer numerical simulation results show that our proposed system could achieve a very small reconstruction error at a very small oversampling ratio.

A new approach to identify obstructive sleep apnea using an optimal orthogonal wavelet filter bank with ECG signals

Abstract Obstructive sleep apnea (OSA) is a severe sleep ailment. It manifests when a person's breathing is interrupted while sleeping due to an inadequate supply of oxygen to the brain and physical body. The detection of OSA at an early stage can provide better mitigation from severe health impairments. An accurate method to detect OSA can improve the quality of life significantly. The computer-aided diagnosis (CAD) system of OSA developed using single-channel, single-signal physiological signals can efficiently cut costs, and can be used even at home. Therefore, a simple and portable OSA-CAD system using single-channel electrocardiogram (ECG) is introduced in this paper. This study proposes the automated identification of ECG signal affected from OSA employing an optimal two-band filter bank (FB) technique. While designing the filter bank, our central objective is to minimize the spectral localization, subjected to exact regeneration and regularity criteria. The identification of OSA using ECG signals is duly based on the newly designed FB. Using the proposed FB, ECG signals were split into wavelet frequency-bands (WFBs). The fuzzy-entropy (FUEN) and the log of signal-energy (LOEN) of WFBs have been employed as the distinguishing features. The 35-fold cross-validation technique was exercised using various classifiers, namely K nearest neighbor (KNN), decision tree (DT), linear discriminant, logistic regression, and support vector machine (SVM) to separate into normal and OSA affected subjects. The proposed model has attained respective highest average accuracy (AVAC), average sensitivity (AVSE), average specificity (AVSP) and F1-score of 90.87%, 92.43% 88.33% and 92.61%. Our proposed model outperformed the existing systems developed using the same database and was found to be more efficient, robust, and easy to use.

A novel automated diagnostic system for classification of myocardial infarction ECG signals using an optimal biorthogonal filter bank

Myocardial infarction (MI), also referred to as heart attack, occurs when there is an interruption of blood flow to parts of the heart, due to the acute rupture of atherosclerotic plaque, which leads to damage of heart muscle. The heart muscle damage produces changes in the recorded surface electrocardiogram (ECG). The identification of MI by visual inspection of the ECG requires expert interpretation, and is difficult as the ECG signal changes associated with MI can be short in duration and low in magnitude. Hence, errors in diagnosis can lead to delay the initiation of appropriate medical treatment. To lessen the burden on doctors, an automated ECG based system can be installed in hospitals to help identify MI changes on ECG. In the proposed study, we develop a single-channel single lead ECG based MI diagnostic system validated using noisy and clean datasets. The raw ECG signals are taken from the Physikalisch-Technische Bundesanstalt database. We design a novel two-band optimal biorthogonal filter bank (FB) for analysis of the ECG signals. We present a method to design a novel class of two-band optimal biorthogonal FB in which not only the product filter but the analysis lowpass filter is also a halfband filter. The filter design problem has been composed as a constrained convex optimization problem in which the objective function is a convex combination of multiple quadratic functions and the regularity and perfect reconstruction conditions are imposed in the form linear equalities. ECG signals are decomposed into six subbands (SBs) using the newly designed wavelet FB. Following to this, discriminating features namely, fuzzy entropy (FE), signal-fractal-dimensions (SFD), and renyi entropy (RE) are computed from all the six SBs. The features are fed to the k-nearest neighbor (KNN). The proposed system yields an accuracy of 99.62% for the noisy dataset and an accuracy of 99.74% for the clean dataset, using 10-fold cross validation (CV) technique. Our MI identification system is robust and highly accurate. It can thus be installed in clinics for detecting MI.

MMSFL-OWFB: A novel class of orthogonal wavelet filters for epileptic seizure detection

The optimal filters with minimal bandwidth are highly desirable in many applications such as communication and biomedical signal processing. In this study, we design optimally frequency localized orthogonal wavelet filters and evaluate their performance using electroencephalogram (EEG) signals for automated detection of the epileptic seizure. The paper presents a novel method for designing optimal orthogonal wavelet filter banks (OWFB) with the objective of minimizing their frequency spreads. The designed wavelet filter also possesses the desired degree of regularity. The regularity condition has been imposed analytically so as to satisfy the constraint accurately. We propose a novel semi-definite programming (SDP) formulation which does not involve any parametrization. The solution of the SDP yields optimal orthogonal wavelet filter for the given length of the filter. We have developed an automated diagnosis system that identifies epileptic seizure EEG signals using the features obtained from the designed minimally mean squared frequency localized (MMSFL) OWFB. We have tested the performance of the proposed model using two independent EEG databases in order to ensure the consistency and robustness of the model. Interestingly, the proposed MMSFL-OWFB feature-based model exhibits ceiling level of performance, with classification accuracy ≥ 99% in classifying seizure (ictal) and seizure-free (non-ictal) EEG signals for both databases. Our developed system can be employed in hospitals and community cares to aid the epileptologists in the accurate diagnosis of seizures.

Application of an optimal class of antisymmetric wavelet filter banks for obstructive sleep apnea diagnosis using ECG signals

Obstructive sleep apnea (OSA) is a sleep disorder caused due to interruption of breathing resulting in insufficient oxygen to the human body and brain. If the OSA is detected and treated at an early stage the possibility of severe health impairment can be mitigated. Therefore, an accurate automated OSA detection system is indispensable. Generally, OSA based computer-aided diagnosis (CAD) system employs multi-channel, multi-signal physiological signals. However, there is a great need for single-channel bio-signal based low-power, a portable OSA-CAD system which can be used at home. In this study, we propose single-channel electrocardiogram (ECG) based OSA-CAD system using a new class of optimal biorthogonal antisymmetric wavelet filter bank (BAWFB). In this class of filter bank, all filters are of even length. The filter bank design problem is transformed into a constrained optimization problem wherein the objective is to minimize either frequency-spread for the given time-spread or time-spread for the given frequency-spread. The optimization problem is formulated as a semi-definite programming (SDP) problem. In the SDP problem, the objective function (time-spread or frequency-spread), constraints of perfect reconstruction (PR) and zero moment (ZM) are incorporated in their time domain matrix formulations. The global solution for SDP is obtained using interior point algorithm. The newly designed BAWFB is used for the classification of OSA using ECG signals taken from the physionet's Apnea-ECG database. The ECG segments of 1 min duration are decomposed into six wavelet subbands (WSBs) by employing the proposed BAWFB. Then, the fuzzy entropy (FE) and log-energy (LE) features are computed from all six WSBs. The FE and LE features are classified into normal and OSA groups using least squares support vector machine (LS-SVM) with 35-fold cross-validation strategy. The proposed OSA detection model achieved the average classification accuracy, sensitivity, specificity and F-score of 90.11%, 90.87% 88.88% and 0.92, respectively. The performance of the model is found to be better than the existing works in detecting OSA using the same database. Thus, the proposed automated OSA detection system is accurate, cost-effective and ready to be tested with a huge database.

Optimum low complexity filter bank for generalized orthogonal frequency division multiplexing

Generalized frequency division multiplexing (GFDM) is one of the multicarrier modulation candidates proposed for the 5th generation of wireless networks. Among GFDM linear receivers, GFDM MMSE receiver achieves the best error performance for multipath fading channels at the cost of high numerical complexity. Hence, the combination of GFDM match filter (MF) receiver and double-side successive interference cancellation (DSIC) method is used instead. However, there is a significant gap between the error performance of GFDM MMSE and DSIC/MF receivers for the case of employing modern channel coding. Recently, we have proposed a new multicarrier scheme based on GFDM architecture called generalized orthogonal frequency division multiplexing (GOFDM). This study derives an optimized cyclic tree-structured perfect reconstruction-quadrature mirror filter (PR-QMF) bank for GOFDM transceiver and then introduces a novel method for implementation of the optimum filter bank in the frequency domain. Employing such a fast and optimum filter bank provides several advantages for GOFDM transceiver. GOFDM transmitter mitigates out-of-band spectrum leak to the level of that of GFDM. In addition, choosing an appropriate configuration of filter bank yields lower peak to average power ratio in transmit signal of GOFDM compared to that of OFDM. On the other hand, while GOFDM MMSE receiver has lower numerical complexity compared with GFDM DSIC/MF receiver, its coded bit error rate curve is close to that of GFDM MMSE receiver. The aforementioned advantages envision GOFDM as a competitive candidate to be employed in the physical layer of new wireless applications.

Optimal Differentiator Filter Banks for PMUs and Their Feasibility Limits

In this paper, we present a very general design approach for optimal linear-phase differentiator filter banks for phasor measurement units based on convex semi-infinite optimization. A detailed presentation of the formulation of both the cost functions and constraints is included for the positive-sequence phasor estimation problem. The design method is extremely powerful and flexible as it allows to control precisely the behavior of the system in terms of the total vector error (TVE), frequency error (FE), and rate of change of frequency error (RFE) metrics for several different scenarios. This feature is extremely useful for tailored designs of these filters for different applications. We also determine numerically the uniform feasibility limits of the system as a function of the filter lengths and, in particular, study the required filter lengths for compliance with the IEEE Standards C37.118.1-2011 and C37.118.1a-2014. It is found that all requirements can be achieved with a significant margin with the exception of the M class FE constraint for interharmonic components, an issue which is also reported in other works and briefly discussed here. Finally, an interesting comparison with the Taylor-Fourier filters is made to illustrate the advantages of our approach.

Split-spectrum processing technique for SNR enhancement of ultrasonic guided wave

Ultrasonic guided wave (UGW) systems are broadly used in several branches of industry where the structural integrity is of concern. In those systems, signal interpretation can often be challenging due to the multi-modal and dispersive propagation of UGWs. This results in degradation of the signals in terms of signal-to-noise ratio (SNR) and spatial resolution. This paper employs the split-spectrum processing (SSP) technique in order to enhance the SNR and spatial resolution of UGW signals using the optimized filter bank parameters in real time scenario for pipe inspection. SSP technique has already been developed for other applications such as conventional ultrasonic testing for SNR enhancement. In this work, an investigation is provided to clarify the sensitivity of SSP performance to the filter bank parameter values for UGWs such as processing bandwidth, filter bandwidth, filter separation and a number of filters. As a result, the optimum values are estimated to significantly improve the SNR and spatial resolution of UGWs. The proposed method is synthetically and experimentally compared with conventional approaches employing different SSP recombination algorithms. The Polarity Thresholding (PT) and PT with Minimization (PTM) algorithms were found to be the best recombination algorithms. They substantially improved the SNR up to 36.9dB and 38.9dB respectively. The outcome of the work presented in this paper paves the way to enhance the reliability of UGW inspections.

벡터 양자화기를 사용한 최적의 부대역 필터 뱅크 구현에 관한 연구

이 논문은 벡터 양자기가 포함된 부대역 코덱의 분석과 설계에서 벡터 양자기를 모델링하는 새로운 방법을 제시해준다. 우리는 각 코드북의 시작점들의 수(N), 각 코드워드의 길이(k), 필터 대역 계수들에 의존하는 부대역 코덱 시스템의 입력과 출력의 평균자승 회복 오차(MSE)를 계산한다. 본 논문은 확률밀도함수로 최적화된 벡터양자기가 존재하는 최적의 M밴드 필터 뱅크 구조는 등가의 스칼라 양자기의 변수들의 적절한 선택으로 구현될 수 있음을 보였다. 특정한 구현 예를 두 개의 다른 필터뱅크 구조인 Paraunitary 필터 뱅크와 Biorthogonal 필터 뱅크를 2채널 경우에 개발하였다. 이 이론적인 결과들은 확장의 Monte Carlo 시뮬레이션으로 확인되었다. This paper provides a new approach for modeling of vector quantizer(VQ) followed by analysis and design of subband codecs with imbedded VQ's. We compute the mean squared reconstruction error(MSE) which depend on N the number of entries in each codebook, k the length of each codeword, and on the filter bank(FB) coefficients in subband codecs. We show that the optimum M-band filter bank structure in presence of pdf-optimized vector quantizer can be designed by a suitable choice of equivalent scalar quantizer parameters. Specific design examples have been developed for two different classes of filter banks, paraunitary and the biorthogonal FB and the 2 channel case. These theoretical results are confirmed by Monte Carlo simulation.

Optimal duration-bandwidth localized antisymmetric biorthogonal wavelet filters

We present a design of a new class of compactly supported antisymmetric biorthogonal wavelet filter banks which have the analysis as well as the synthesis filters of even-length. Here, the analysis and the synthesis filters are designed to have minimum joint duration-bandwidth localization (JDBL). The design of filters has been formulated as a direct time-domain linearly constrained eigenvalue problem that does not involve any parametrization and iterations. The optimal analysis and synthesis filters have been obtained as the eigenvectors of the positive definite matrices. The closed form analytic expression for the objective function has been presented. The perfect reconstruction and regularity conditions have been incorporated in the design by employing time-domain matrix characterization. The method can control duration and bandwidth localizations of the analysis and synthesis filters, independently. A few design examples have been presented and compared with previous works. The performance of the optimal filter banks designed by employing the proposed method has been evaluated in image coding and signal denoising applications.

Orthogonal Design of Cyclic Block Filtered Multitone Modulation

The orthogonal design of a cyclic block filtered multititone modulation (CB-FMT) system is addressed. CB-FMT is a filter bank modulation scheme that uses frequency confined prototype pulses, similarly to FMT. Differently from FMT, where the linear convolution is used, the cyclic convolution is exploited in CB-FMT. This allows to efficiently implement the system via a concatenation of discrete Fourier transforms. The necessary and sufficient orthogonality conditions are derived in time domain and frequency domain. Then, these conditions are expressed in matrix form, and the prototype pulse coefficients are parameterized with hyper-spherical coordinates. The effect of a linear time-variant transmission medium is discussed. In such a scenario, the optimal filter bank orthogonal design is considered with the objective of maximizing either the in-band-to-out-band sub-channel energy ratio or the achievable rate. Numerical results and comparisons show the performance improvements attainable with several designed optimal pulses also with respect to the use of the baseline root-raised-cosine pulse.

Design of Time–Frequency Optimal Three-Band Wavelet Filter Banks with Unit Sobolev Regularity Using Frequency Domain Sampling

In this paper, we design three-band time---frequency-localized orthogonal wavelet filter banks having single vanishing moment. We propose new expressions to compute mean and variances in time and frequency from the samples of the Fourier transform of the asymmetric band-pass compactly supported wavelet functions. We determine discrete-time filter of length eight that generates the time---frequency optimal time-limited scaling and wavelet functions using cascade algorithm. Time---frequency product (TFP) of a function is defined as the product of its time variance and frequency variance. The TFP of the designed functions is close to 0.25 with unit Sobolev regularity. Three-band filter banks are designed by minimizing a weighted combination of TFPs of wavelets and scaling functions. Interestingly, empirical results show that time---frequency optimal, filter banks of length nine, designed with the proposed methodology, have unit Sobolev regularity, which is maximum achievable with single vanishing moment. Design examples for length six and length nine filter banks are given to demonstrate the effectiveness of the proposed design methodology.

Design of Time–Frequency Localized Filter Banks: Transforming Non-convex Problem into Convex Via Semidefinite Relaxation Technique

We present a method for designing optimal biorthogonal wavelet filter banks (FBs). Joint time---frequency localization of the filters has been chosen as the optimality criterion. The design of filter banks has been cast as a constrained optimization problem. We design the filter either with the objective of minimizing its frequency spread (variance) subject to the constraint of prescribed time spread or with the objective of minimizing the time spread subject to the fixed frequency spread. The optimization problems considered are inherently non-convex quadratic constrained optimization problems. The non-convex optimization problems have been transformed into convex semidefinite programs (SDPs) employing the semidefinite relaxation technique. The regularity constraints have also been incorporated along with perfect reconstruction constraints in the optimization problem. In certain cases, the relaxed SDPs are found to be tight. The zero duality gap leads to the global optimal solutions. The design examples demonstrate that reasonably smooth wavelets can be designed from the proposed filter banks. The optimal filter banks have been compared with popular filter banks such as Cohen---Daubechies---Feauveau biorthogonal wavelet FBs, time---frequency optimized half-band pair FBs and maximally flat half-band pair FBs. The performance of optimal filter banks has been found better in terms of joint time---frequency localization.

A novel analytical integer optimization method for wavelet based subband coding

In subband coding (SBC), the original signal is decomposed into some frequency subbands and then the total available number of bits is divided between different subbands of the signal. In the most of existing SBC methods, the number of allocated bits can be real and negative, while in practice the number of bits must be integer and nonnegative. In this paper an analytical solution is derived for subband coding with optimum nonnegative integer bit allocation and multi-resolution filter bank (including wavelet filter bank). The analytical solution is applicable for either non-uniform or uniform SBC. A modified discrete bisection algorithm is also proposed which can reduce the computational complexity of searching in a group of discrete functions. The computational complexity of proposed method is lower than the complexity of integer optimization algorithms which are applicable to SBC. Compared to the common SBC algorithms with real-valued bit allocation (in which the number of bits should be rounded), the proposed method has much less quantization error.

Meta-heuristic evolutionary algorithms for the design of optimal multiplier-less recombination filter banks

This paper proposes a design for multiplier-less recombination non-uniform filter banks (RNUFBs) optimized using meta-heuristic algorithms. The structure consists of an M-channel uniform filter bank, with some channels combined by the synthesis filters of a transmultiplexer (TMUX), yielding non-uniform sub-bands. When any structure is realized in hardware, it is necessary to have low power consumption and a small chip area. These can be achieved by replacing the multipliers with shifters and adders. Once the continuous coefficient recombination non-uniform filter bank is designed, the coefficients are converted to the canonic-signed-digit (CSD) space to make the design multiplier-less, so as to reduce the complexity of the hardware implementation. To reduce the number of adders and shifters in the multiplier-less implementation, the filter coefficients are rounded with a restricted number of signed power-of-two (SPT) terms, which may cause degradation in the performance of the RNUFBs. To improve the performance of the CSD rounded filters and filter bank, meta-heuristic algorithms such as the artificial bee colony (ABC) algorithm, harmony search algorithm (HSA) and gravitational search algorithm (GSA) are deployed. Of these meta-heuristic algorithms, GSA is found to give the best performance. The method proposed in this paper results in non-uniform filter banks with rational sampling factors which are multiplier-less and have linear-phase and near-perfect-reconstruction.

Wideband Waveform and Receiver Filter Bank Design for Clutter Suppression

Modern wideband radar systems with long integration time, equipped with arbitrarily waveform generators, raise a demand for advanced signaling transmission schemes. In this paper, we propose two algorithms to select transmit waveforms and receiver filters. The techniques are based on a clutter suppression criterion. For the first algorithm, we employ an optimized filter bank, and for the second algorithm, we use a matched filter bank. Clutter suppression is achieved by minimizing the correlation between receiver filters and interfering clutter echoes. The algorithm, for the optimized filter bank, is extended to adapt the transmission scheme and receiver filters to a time-evolving scenario. Adaptation parameters are based on estimates of a clutter map and detected target characteristics. To estimate the clutter map we propose a Kalman filter, whereas target parameters are calculated using a least-squares fit to data. The efficiency of the algorithms and the adaptation scheme are visualized through a numerical simulation.

Fuzzy decision fusion of complementary experts based on evolutionary cepstral coefficients for phoneme recognition

Optimal representation of acoustic features is an ongoing challenge in automatic speech recognition research. As an initial step toward this purpose, optimization of filterbanks for the cepstral coefficient using evolutionary optimization methods is proposed in some approaches. However, the large number of optimization parameters required by a filterbank makes it difficult to guarantee that an individual optimized filterbank can provide the best representation for phoneme classification. Moreover, in many cases, a number of potential solutions are obtained. Each solution presents discrimination between specific groups of phonemes. In other words, each filterbank has its own particular advantage. Therefore, the aggregation of the discriminative information provided by filterbanks is demanding challenging task. In this study, the optimization of a number of complementary filterbanks is considered to provide a different representation of speech signals for phoneme classification using the hidden Markov model (HMM). Fuzzy information fusion is used to aggregate the decisions provided by HMMs. Fuzzy theory can effectively handle the uncertainties of classifiers trained with different representations of speech data. In this study, the output of the HMM classifiers of each expert is fused using a fuzzy decision fusion scheme. The decision fusion employed a global and local confidence measurement to formulate the reliability of each classifier based on both the global and local context when making overall decisions. Experiments were conducted based on clean and noisy phonetic samples. The proposed method outperformed conventional Mel frequency cepstral coefficients under both conditions in terms of overall phoneme classification accuracy. The fuzzy fusion scheme was shown to be capable of the aggregation of complementary information provided by each filterbank.