Showing papers on "Time–frequency analysis published in 1990"
TL;DR: Experimental results on simulated data and real speech show the advantages of the GTFRs with the cone-shaped kernels through comparisons to the spectrogram and the pseudo-Wigner distribution.
Abstract: Generalized time-frequency representations (GTFRs) which use cone-shaped kernels for nonstationary signal analysis are presented. The cone-shaped kernels are formulated for the GTFRs to produce good resolution simultaneously in time and frequency. Specifically, for a GTFR with a cone-shaped kernel, finite time support is maintained in the time dimension along with an enhanced spectrum in the frequency dimension, and the cross-terms are smoothed out. Experimental results on simulated data and real speech show the advantages of the GTFRs with the cone-shaped kernels through comparisons to the spectrogram and the pseudo-Wigner distribution. >
461 citations
TL;DR: The Q distributions, a modified W-V representation that is related to the wide-band ambiguity function by an integral transform, is defined and properties of the Q distribution indicate that it may prove useful for detection and parameter estimation as well as for tomographic measurement of wideband scattering functions with relatively few transmitted waveforms.
Abstract: The Wigner-Ville (W-V) distribution is a time-frequency representation that yields a highly accurate estimate of instantaneous frequency. It is related to the narrow-band ambiguity function by an internal transform, and it can be used in a variety of detection and estimation problems. A spectrogram constructed with constant bandwidth filters can be obtained by convolving two W-V distributions or by forming a magnitude-squared narrow-band cross-ambiguity function. The wide-band ambiguity function represents the Doppler effect with dilation or compression rather than with frequency shift as in the narrow-band approximation. The Q distributions, a modified W-V representation that is related to the wide-band ambiguity function by an integral transform, is defined. A spectrogram constructed with proportional-bandwidth or constant-Q filters can be obtained by a convolution-like operation involving two Q distributions or by forming a magnitude-squared wide-band cross-ambiguity function. The Q distribution is thus a wide-band version of the W-V distribution. Properties of the Q distribution indicate that it may prove useful for detection and parameter estimation as well as for tomographic measurement of wideband scattering functions with relatively few transmitted waveforms. >
68 citations
01 Nov 1990
TL;DR: This paper compares the two transforms and makes the case that the Gabor representation can often be more compact, and may require substantially less computation and storage in some applications.
Abstract: Effective signal detection and feature extraction in noisy environments generally depend on exploiting some knowledge of the signal. The short-time Fourier transform and the Gabor transform are two methods that exploit signal envelope information. This paper compares the two transforms and makes the case that the Gabor representation can often be more compact, and may require substantially less computation and storage in some applications. There is a sense in which the Gabor achieves a preferential trade of SNR for resolution, and because of this, one can also expect better signal recognition and feature reconstructions from the Gabor transform in the presence of noise.© (1990) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
16 citations
TL;DR: The classical theory of speech production proves the validity of the EWSM parameters; their modifications yield well-localized time-frequency transformations, including frequency compression/expansion, pitch, formant and noise modification.
15 citations
TL;DR: It is shown that a wide class of time-frequency distributions gives the power spectrum for the case of a wide-sense stationary random signal.
Abstract: It is shown that a wide class of time-frequency distributions gives the power spectrum for the case of a wide-sense stationary random signal. A simple criterion on the Cohen (1966) kernel for determining when a bilinear time-frequency distribution gives the power spectrum of the signal is derived. >
12 citations
03 Apr 1990
TL;DR: It is shown that the GTFR technique offers a representation which is less sensitive to threshold settings and is thus more robust and potentially more important shorter-term advantages are illustrated and quantified as more accurate locations of bursts in the time-frequency distribution.
Abstract: A nonstationary analysis technique is applied to speech, with encouraging results. This technique makes use of a generalized time-frequency representation (GTFR) which uses a kernel that allows finite-time support while suppressing interference terms. This kernel has a cone shape in the (t, tau ) plane, where t is time of a signal and tau is an autocorrelationlike lag. The processing thus allows time and frequency resolution equivalent to the Wigner distribution but does not have significant interference terms. It is shown that the effect of this distribution on the long-term (1.5-s) display of speech is a visible enhancement of formant tracks. Potentially more important shorter-term (30-ms or less) advantages are illustrated and quantified as more accurate locations of bursts in the time-frequency distribution. It is shown that the GTFR technique offers a representation which is less sensitive to threshold settings and is thus more robust. >
12 citations
01 Nov 1990
TL;DR: The purpose of this paper is to provide a brief overview of linear wavelet techniques and bilinear time-scale methods and to put them insome common perspective with existing Signal Processing tools (Gabor decompositions, constant-Q analysis, quadrature mirror filters, wideband ambiguity functions, time-frequency energy distributions.)
Abstract: In a very recent past, new techniques, referred to as time-scale methods and making use of the so-called wavelet transform, have been proposed for the analysis of nonstationary or time-varying signals. They arebasically devoted to the description of signal time evolutions at different observation scales : this is achieved byusing shifted and dilated versions of some elementary analyzing waveform along the time axis. The purpose ofthis paper is twofold : it is intended 1) to provide a brief overview of linear wavelet techniques (continuous anddiscrete transforms) and bilinear time-scale methods (time-scale energy distributions), and 2) to put them insome common perspective with existing Signal Processing tools (Gabor decompositions, constant-Q analysis,quadrature mirror filters, wideband ambiguity functions, time-frequency energy distributions.) Existing orpotentially relevant applications are also pointed out. 1. Introduction Whereas most of physical phenomena give rise to nonstationary signals (with, moreover, the richestpart of their information conveyed by nonstationarities), most widely spread signal processing toolsdo not encompass any time-dependence and, therefore, are not able to capture local properties ofsignals. This is especially true for Fourier-based methods, which basically rely on the physicallyunrealisable concept of infinite wave. The search for more realistic signal decompositions has firstresulted in time-frequency methods, whose aim was primarily to make spectral analysis become time-dependent. In a recent past, it was realized that nonstationarities could also be handled by consideringthe time evolution of a parameter which differs from frequency, namely a scale parameter.The search for time-scale signal descriptions has been motivated not only by the importantnumber of physical situations in which scaling behaviors are observed, but also by theoretical andalgorithmic reasons. It has led to considerable work regarding applied mathematics and theoreticalphysics, as well as signal processing, the corresponding body of knowledge being referred to as the
7 citations
04 Dec 1990
TL;DR: In this article, the phase behavior of the wavelet transform has been studied for the characterization of elastic targets immersed in a fluid and subjected to an acoustic impulse, and an algorithm based on the behaviour of the phase of the transform was developed.
Abstract: The characterization of elastic targets immersed in a fluid and subjected to an acoustic impulse is discussed. The wavelet transform has been chosen for its particular properties, such as linearity and local analysis at Delta f/f=c/sup ste/. An algorithm based on the behavior of the phase of the transform has been developed. It allows the extraction of modulation laws (related to the dispersion law of the phase velocity), even for close echoes. In the case of spherical elastic shells, the method was used on both experimental and computer-generated signals, and the good relation between theoretical and experimental results is presented. >
7 citations
03 Apr 1990
TL;DR: In this article, a method based on an iterative procedure incorporating time-frequency distributions (TFDs) is proposed for instantaneous frequency (IF) estimation of rapidly time-varying signals, which is then applied to the problem of adaptively detecting a transient of unknown waveshape in white Gaussian noise.
Abstract: A method based on an iterative procedure incorporating time-frequency distributions (TFDs) is proposed for instantaneous frequency (IF) estimation of rapidly time-varying signals. This method is then applied to the problem of adaptively detecting a transient of unknown waveshape in white Gaussian noise. For this type of adaptive detection, the signal's time-frequency representation is first estimated, and this estimate is used as the true representation in a time-frequency correlator detector. The IF is assumed to be a critical feature of the transient. Accordingly, the IF is first estimated, and this estimate is then used as an aid to estimate the time-frequency representation of the true signal. This representation is then correlated with the time-frequency representation of the observed signal to provide the appropriate detection statistics. >
7 citations
01 Nov 1990
TL;DR: In this article, the scale dependent wavelet analysis is replaced by rotated basis functions such that an image is represented by a superposition of translated differently rotated versions of the same basis function.
Abstract: Conventional scale dependent wavelet analysis represents a signal or iniage as a superposition of translated differently scaled versions of the same basis function. When the basis function for time series analysis is a chirp with linear frequency modulation a scale dependent wavelet representation is equivalent to a sequence of projections of the signal timefrequency distribution along differently rotated lines and reconstruction of the signal from its chirped wavelet representation is analogous to tomographic reconstruction from time frequency projections. The same analogy applies in two dimensions if scaled basis functions are replaced by rotated ones such that an image is represented by a superposition of translated differently rotated versions of the same basis function. For rotation dependent wavelet analysis basis functions consisting of very long line segments yield a tomographic representation while shorter line segments yield a line segment image representation as in the primate visual cortex. Applications include binocular robot vision and synthetic aperture radar.© (1990) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
21 May 1990
TL;DR: The PT approach is shown to provide a more natural data compression mechanism than the hybrid approach for reducing implementation complexity, while yielding satisfactory performance.
Abstract: A comparative analysis of the implementation reduction mechanisms present in a predictive transform (PT) estimation scheme and in a conventional time-frequency or hybrid estimation approach inspired by hybrid coding is performed. This comparative study focuses on the estimation dual of the hybrid versus predictive transform coding problem, and thus serves to further illustrate the ability of the PT approach to unify seemingly disjoint areas of signal processing. A Kalman-type estimation example, based on a previously considered NTSC image reconstruction problem, is used to illustrate the above ideas. The PT approach is shown to provide a more natural data compression mechanism than the hybrid approach for reducing implementation complexity, while yielding satisfactory performance. >
01 Apr 1990
TL;DR: In this paper, a singular-valued-decomposition-based algorithmic procedure which significantly increases the signal-noise ratio (SNR) and preserves the shape of the time-frequency signal is presented.
Abstract: A singular-valued-decomposition-based algorithmic procedure which significantly increases the signal-noise ratio (SNR) and preserves the shape of the time-frequency signal is presented. This algorithm is effective for both small and large time-bandwidth product signals. It is based on a set of Wigner distribution properties for such signals and is formulated in a time-frequency signal subspace. The results from applying the algorithm to the task of recovering wideband chirp signals in a low-SNR environment are examined. >
05 Nov 1990
TL;DR: How a combined time-frequency description offers a powerful approach for the study of acoustic scattcring is described and the advantages and disadvantages of various timefrequency distributions are discussed.
Abstract: We describe how a combined time-frequency description offers a powerful approach for the study of acoustic scattcring, axid we discuss the advantages and disadvantages of various timefrequency distributions. Results of scattering experiments performed in an acoustic tank are presented to illustrate tlic uqe of these methods.
01 Apr 1990
TL;DR: A time evolving spectral pseudosemblance, or coherency, function is proposed where the expectation of the Wigner distribution is shown to be equal to the instantaneous power spectral density of a nonstationary signal.
Abstract: A time evolving spectral pseudosemblance, or coherency, function is proposed. The function relies on auto- and cross-Wigner distribution analysis where the expectation of the Wigner distribution is shown to be equal to the instantaneous power spectral density of a nonstationary signal. Signal returns are modeled as the output of a time-variant random filter through which the time-evolutionary nature of the function is shown to offer a natural extension to the magnitude squared coherency function for stationary signals. >