By Topic

Acoustics, Speech and Signal Processing, IEEE Transactions on

Issue 6 • Date December 1986

Filter Results

Displaying Results 1 - 25 of 42
  • [Front cover and table of contents]

    Page(s): 0
    Save to Project icon | Request Permissions | PDF file iconPDF (324 KB)  
    Freely Available from IEEE
  • Full text access may be available. Click article title to sign in or learn about subscription options.
  • [Back cover]

    Page(s): c4
    Save to Project icon | Request Permissions | PDF file iconPDF (3369 KB)  
    Freely Available from IEEE
  • WMMSE design of digital filter banks with specified composite response

    Page(s): 1529 - 1541
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1408 KB)  

    A new method for designing uniform and nonuniform digital filter banks with a specified composite response is presented. The composite response of the filter bank can be met either exactly or to within a given tolerance. We focus on filter banks in which the individual filters are finite impulse response (FIR) digital filters of possibly nonequal length, although the new method is applicable even to more general structures as well. The new method minimizes the weighted sum of the mean square errors in the response of the individual filters, subject to the composite response specifications. Sufficient conditions for either real, ness or phase linearity of the optimal individual filters are presented. The new weighted minimum mean square error (WMMSE) design method is interpreted from a statistical viewpoint as a maximization of the harmonic mean of the output signal-to-noise ratio (SNR) of the individual filters. The complexity of the new method is analyzed, and the design process is demonstrated via a design example. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On an adaptive noise cancellation application for radar

    Page(s): 1654 - 1655
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (208 KB)  

    The purpose of this correspondence is to introduce a novel application of adaptive noise cancellation related to a class of radar signals. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive equalizer using finite-bit power-of-two quantizer

    Page(s): 1603 - 1611
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (744 KB)  

    The stochastic gradient algorithm using a simplified arithmetic is analyzed in this paper. A power-of-two quantizer is used for the input of the multiplier to reduce the multiplication to at most a simple shift. In spite of its simple implementation, the performance is shown to be comparable to the classical LMS algorithm. A linearized approximation to the quantizer is first derived, followed by the analysis of an exact nonlinear model. The derivation is based on the Gaussian assumption, and the effects of removing the Gaussian assumption are later considered. The roundoff error due to the finite-bit computation is calculated. Computer simulation results are provided to support the analysis. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Speech transformations based on a sinusoidal representation

    Page(s): 1449 - 1464
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1528 KB)  

    In this paper a new speech analysis/synthesis technique is presented which provides the basis for a general class of speech transformations including time-scale modification, frequency scaling, and pitch modification. These modifications can be performed with a time-varying change, permitting continuous adjustment of a speaker's fundamental frequency and rate of articulation. The method is based on a sinusoidal representation of the speech production mechanism which has been shown to produce synthetic speech that preserves the wave-form shape and is perceptually indistinguishable from the original. Although the analysis/synthesis system was originally designed for single-speaker signals, it is also capable of recovering and modifying nonspeech signals such as music, multiple speakers, marine biologic sounds, and speakers in the presence of interferences such as noise and musical backgrounds. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Rotational search methods for adaptive Pisarenko harmonic retrieval

    Page(s): 1550 - 1565
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1272 KB)  

    Two adaptation algorithms for adaptive Pisarenko harmonic retrieval are described. They are derived by considering the associated minimum eigenvalue problem as an optimization problem which seeks the minimum of a quadratic cost function given a hyperspherical constraint. An iterative search procedure is used in which each search path is constrained to lie on the unit hypersphere. Computational complexity per iteration is approximately one-third that of previous adaptive PHR algorithms. Simulations reveal that at low SNR the trial eigenvector can converge to the true minimum eigenvector of the sample covariance matrix, long before this matrix is a good estimate of the true covariance matrix. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A self-orthogonalizing efficient block adaptive filter

    Page(s): 1573 - 1582
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1088 KB)  

    This paper deals with the development of a unique self-orthogonalizing block adaptive filter (SOBAF) algorithm that yields efficient finite impulse response (FIR) adaptive filter structures. Computationally, the SOBAF is shown to be superior to the least mean squares (LMS) algorithm. The consistent convergence performance which it provides lies between that of the LMS and the recursive least squares (RLS) algorithm, but, unlike the LMS, is virtually independent of input statistics. The block nature of the SOBAF exploits the use of efficient circular convolution algorithms such as the FFT, the rectangular transform (RT), the Fermat number transform (FNT), and the fast polynomial transform (FPT). In performance, the SOBAF achieves the mean squared error (MSE) convergence of a self-orthogonalizing structure, that is, the adaptive filter converges under any input conditions, at the same rate as an LMS algorithm would under white input conditions. Furthermore, the selection of the step size for the SOBAF is straightforward as the range and the optimum value of the step size are independent of the input statistics. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A rearranged DFT algorithm requiring N2/6 multiplications

    Page(s): 1658 - 1659
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (256 KB)  

    We consider the problem of computing the DFT and present two reductions over the standard formula. In the special case of an N-point sequence with N = 2l, the number of multiplications per output point required by this algorithm is, at most, N/4 - 1 and, on the average, N/6 - 1. Each output point requires no more than N - 1 additions. In applications requiring only some of the output points, a computational savings over the standard (FFT) techniques may be achieved. Furthermore, we argue that in a certain sense these reductions are optimal. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Shape-gain matrix quantizers for LPC speech

    Page(s): 1427 - 1439
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1312 KB)  

    It has been recently demonstrated that the principles of vector quantization for LPC speech can be simply extended to encompass matrices of LPC vectors with significant savings in bit rate. Unfortunately, however, such locally optimal matrix quantizers have prohibitively high complexity and memory requirements when implemented in a speech vocoder at bit rates giving acceptable quality speech. One approach to solving the problem is to separately code gain and shape in the matrix quantizer. This paper generalizes the principles of shape-gain vector quantizer design for LPC speech to matrix quantization and investigates the properties of the resulting quantizers. In particular, we present a design which combines shape matrices consisting of N shape vectors with K-dimensional gain vectors, where N and K are small integers, in practice, with K \geq N . Experimental results show that with K, N \geq 3 , significant reductions in bit rate over locally optimal vector quantizers are obtained for comparable performance. Simulations indicate that a shape-gain matrix quantizer, using a 10 bit shape codebook and an 8 bit codebook with K = N = 3 operating at 6 bits/frame for the LPC model, gives speech quality comparable to a locally optimal vector quantizer at 9 bits/frame. The matrix quantizer has somewhat greater than 5.7 times the memory requirement of the above vector quantizer, but less than 2.1 times the complexity. Subjective tests show that the speech from this matrix quantizer is intelligible to native speakers of English. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Zero-tracking adaptive filters

    Page(s): 1566 - 1572
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (712 KB)  

    A new type of adaptive filter is proposed which can directly estimate and track its own zeros. The adaptation algorithm adapts the zeros of the filter and hence, indirectly, the filter coefficients. To first order in the adaptation parameter, the new algorithm is equivalent to the usual LMS algorithm, and thus it shares the same convergence properties with the latter. The cases of adaptive prediction, the adaptive Pisarenko method, and adaptive point-source location are discussed in detail. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A convergence analysis of a passive underwater tracking system with nonlinear feedback

    Page(s): 1401 - 1409
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (832 KB)  

    The basic signal processing structure of a new passive underwater tracking system incorporating nonlinear feedback [1], [2] is modeled, and its ability to converge to an unbiased estimate of target range is examined. Target measurements are derived using difference of arrival times between passive sensor systems geometrically separated. The range tracking system utilizes a nonlinear signal processor to first linearize and invert the noisy time delay measurements. This eliminates the need for extended Kalman filtering techniques and allows the use of a more basic-type state estimator [1]. However, the processed measurements now contain both nonstationary and non-Gaussian measurement errors. To help compensate for these effects, the tracking system incorporates nonlinear feedback from output to input, in order to help maintain a zero-mean measurement error process. Thus, a theoretical investigation is necessary to examine overall tracking system convergence after an initial target detection and/or target maneuver has occurred. The convergence analysis is performed using two separate tracking models. The first model is a scalar first-order low-pass filter. The second model is a vector Kalman-type state estimator. Although the estimator is linear, the overall tracking system is nonlinear due to the non-linear bias removal feedback and the data linearization system. This inherent complexity requires the convergence analysis to be both analytic and make extensive use of computer simulation analysis. Results show that each system converges, but with a small bias that is both geometry and signal-to-noise ratio dependent. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Sequences with positive semidefinite Fourier transforms

    Page(s): 1502 - 1510
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (816 KB)  

    A sequence is said to be positive if its Fourier transform exclusively takes on real nonnegative values as a function of frequency. Positive sequences play a prominent role in contemporary signal processing and system theory. To illustrate this point, it is well known that the factorization theorem is extensively used in studies related to wide-sense stationary random signals and linear systems. The ability to appropriately factorize a Fourier transform is contingent on that transform being positive semidefinite. This paper is partially tutorial in that some fundamental positive sequence properties found in dispersed sources are first reviewed. This is followed by the development of several new properties. These properties are in turn used to develop an efficient algorithm for finding that positive sequence which lies closest to a given nonpositive sequence in the least-squares error sense. Interest in this approximation problem arises from the fact that although a given sequence may be theoretically positive, practical considerations often result in its realization being nonpositive. For instance, unbiased autocorrelation lag estimates can lead to nonpositive spectral density function estimates. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Approximate realization algorithms for truncated impulse response data

    Page(s): 1583 - 1588
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (464 KB)  

    The objectives of this paper are: 1) to present a novel derivation of an approximate realization algorithm using singular value decomposition proposed by Kung; 2) to explain the difference between two versions of this algorithm which have been used interchangeably in the literature; and 3) to introduce a new version of this algorithm which exhibits the accuracy of the original algorithm and the theoretical guarantee of stability. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The computation of line spectral frequencies using Chebyshev polynomials

    Page(s): 1419 - 1426
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (832 KB)  

    Line spectral frequencies provide an alternate parameterization of the analysis and synthesis filters used in linear predictive coding (LPC) of speech. In this paper, a new method of converting between the direct form predictor coefficients and line spectral frequencies is presented. The system polynomial for the analysis filter is converted to two even-order symmetric polynomial with interlacing roots on the unit circle. The line spectral frequencies are given by the positions of the roots of these two auxiliary polynomials. The response of each of these polynomials on the unit circle is expressed as a series expansion in Chebyshev polynomials. The line spectral frequencies are found using an iterative root finding algorithm which searches for real roots of a real function. The algorithm developed is simple in structure and is designed to constrain the maximum number of evaluations of the series expansions. The method is highly accurate and can be used in a form that avoids the storage of trigonometric tables or the computation of trigonometric functions. The reconversion of line spectral frequencies to predictor coefficients uses an efficient algorithm derived by expressing the root factors as an expansion in Chebyshev polynomials. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fast Hadamard transform based on a simple matrix factorization

    Page(s): 1666 - 1667
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (224 KB)  

    This correspondence presents a simple factorization of the Hadamard matrix which is used to develop a fast algorithm for the Hadamard transform. The matrix decomposition is of the form of the Kronecker products of identity matrices and successively lower order Hadamard matrices. This decomposition leads very clearly to a sparse-matrix factorization of the Hadamard matrix. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The relation between maximum likelihood estimation of structured covariance matrices and periodograms

    Page(s): 1661 - 1662
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (272 KB)  

    A generalized Burg technique has been developed recently by Burg, Luenberger, and Wegner for maximum likelihood estimation of structured covariance matrices. In this correspondence, the unique solution for the positive definite estimate over a class of nonnegative definite, symmetric matrices with known eigenvectors is presented. This solution coincides with the Karhunen-Loève expansion, and for the class of circulant matrices can be interpreted in terms of periodograms. For stationary processes and infinitely large sample size, it is shown that the sequence of optimal covariance matrices among the class of circulant matrices is asymptotically equivalent to the sequence of true covariance matrices as the observation length approaches infinity. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Bias in the cross spectrum and time delay estimates due to misalignment

    Page(s): 1662 - 1665
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (440 KB)  

    The effect of misalignment in the weighted, overlapped segment averaging method used for estimating time delay is investigated. With nonwhite input, it is shown that the phase of the cross spectrum and the location of the peak in the cross correlator are biased. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On the relation between the maximum entropy probability density function and the autoregressive model

    Page(s): 1659 - 1661
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (368 KB)  

    The problem of maximizing the entropy of an n-variate random vector subject to constraints on the first p + 1 autocovariance terms is examined. It is shown that the maximum is achieved by the Gaussian autoregressive process of order p satisfying the autocovariance constraints. This solution provides Burg's theorem about the maximum entropy spectral density as a special case. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On averaging burg spectral estimators for segments

    Page(s): 1473 - 1484
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1120 KB)  

    The Burg spectral estimator (BSE) exhibits better peak resolution than conventional linear spectral estimators, particularly for short data records. Based on this property, the quality of the BSE is investigated with the available data record segmented and the relevant parameters or functions associated with each segment averaged. Averaging of autoregressive coefficients, reflection coefficients, or spectral density functions is used with the BSE, and the corresponding performances are studied. Approximate expressions for the mean and variance of these modified Burg spectral estimators are derived. The variance of the estimation errors associated with the modified power spectral density estimators is compared to the theoretical Cramer-Rao lower bound. It is observed from the results that averaging of reflection or autoregressive coefficients has almost no effect on bias and variance of the corresponding estimators. Averaging of reflection coefficients is most robust to segmenting, and is therefore recommended for applications using fixed hardware implementations of the Burg algorithm. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A comparison of numerator estimators for ARMA spectra

    Page(s): 1668 - 1671
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (464 KB)  

    This correspondence investigates the problems of estimating the numerator spectrum corresponding to an ARMA time series model once the denominator spectrum (i.e., the AR coefficients) has been estimated. A general form for an estimator of the numerator spectral (NS) coefficients is developed first. Six NS estimators from the recent literature are then compared by fitting them into this general framework and extracting their particular characteristics. It is shown that some methods are special cases of other methods, and that several of these methods are asymptotically equivalent. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On the number of signals resolvable by a uniform linear array

    Page(s): 1361 - 1375
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1448 KB)  

    An algebraic limitation on the maximum number of directions of arrival of plane waves that can be resolved by a uniform linear sensor array is studied. Achievable lower and upper bounds are derived on that number as a function of the number of elements in the array, number of snapshots, and the rank of the source sample-correlation matrix. The signals are assumed narrow-band and of identical and known center frequency. The results are also applicable in the coherent signal case and when directions of arrival are estimated from few snapshots. While in the multiple snapshot case the lower bounds coincide with known asymptotic results, the upper bounds indicate the potential for resolving more signals than by present methods of array processing. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Form-invariant linear filtering: Theory and applications

    Page(s): 1612 - 1628
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (2840 KB)  

    The form-invariant filters are, by definition, those shift-variant filters such that their output, for any given input, turns out to be linearly scaled (implying that its "form" does not change) whenever the input is linearly scaled. In this paper the most general classes of 1- D and 2-D linear form-invariant filters are derived and their properties are discussed, together with their implementation techniques. Two main implementation approaches are considered: one based on the Mellin transform, the other on a combination of coordinate mappings and shift-invariant filtering. The paper also discusses the related works of other authors covering quite different fields such as optical pattern recognition, image restoration and image reconstruction from projections, radar signal processing, etc. It is shown that the mathematics of form-invariant filtering provides a common framework, if not a powerful unified approach, to the many signal processing techniques spread in the above-mentioned works and apparently different application areas. The paper ends with a processing example showing the usefulness of form-invariant filtering in a pattern recognition problem, that is, in the area where the most promising applications of such a filtering are foreseen. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive detection of transient signals

    Page(s): 1410 - 1418
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (784 KB)  

    The paper discusses the problem of detecting transient signals of unknown waveforms in white Gaussian noise. The signals are modeled as impulse responses of rational transfer functions with unknown parameters. A modified generalized likelihood ratio test (MGLRT) is proposed and its statistical properties are analyzed for both known and unknown noise variances. The MGLRT involves constrained maximum likelihood estimation of the signal parameters. The performance of the MGLRT is compared to that of an optimal matched filter and an energy detector, for some test cases. Also, the theoretical distributions of the likelihood ratios under H0and H1are compared to experimental distributions obtained by Monte Carlo simulations. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.

Aims & Scope

This Transactions ceased production in 1990. The current retitled publication is IEEE Transactions on Signal Processing.

Full Aims & Scope