The authors propose an algorithm in which multi-mode coding is a transparent adaptation of multi-stage VXC (vector excitation coding) requiring almost zero additional computation. In the process, it is shown how the c...
详细信息
The authors propose an algorithm in which multi-mode coding is a transparent adaptation of multi-stage VXC (vector excitation coding) requiring almost zero additional computation. In the process, it is shown how the cost of the jointly optimized multi-stage VXC can be minimized using a backward-transform implementation of the perceptually orthogonal VXC (POVXC) algorithm previously proposed by the authors. It is shown that the performance of such a coder, while not as good as that of a normal multi-mode coder, is noticeably better than that of a normal POVXC.< >
The authors introduce two techniques for improving low-bit-rate CELP (code excited linear prediction) speech coders. The sparse-delta stochastic codebook is a recursive codebook design which can be searched using roug...
详细信息
The authors introduce two techniques for improving low-bit-rate CELP (code excited linear prediction) speech coders. The sparse-delta stochastic codebook is a recursive codebook design which can be searched using roughly 5% of the computational load required to search a full Gaussian codebook. Pitch sharpening is a class of algorithms which attempt to improve the perceptual quality of CELP by limiting the feedback of low-amplitude noiselike information to the adaptive codebook. Simulation results are presented for sparse-delta, ternary sparse-delta, and simplified-search sparse-delta coders, and for reduced-gain and sparse-adaptive-codebook pitch sharpening algorithms.< >
In this paper we define a class of continuous-time discrete event dynamic systems (DEDS) with two types of discrete-valued input and output signals: condition signals and event signals called Condition/Event systems (...
详细信息
In this paper we define a class of continuous-time discrete event dynamic systems (DEDS) with two types of discrete-valued input and output signals: condition signals and event signals called Condition/Event systems (C/E systems). We formally define C/E systems and models of C/E systems that are based on an extension of Petri nets (C/E PNs). We then introduce equivalent C/E PN models for C/E systems resulting from cascade and feedback interconnections via examples. The algorithms that compute these equivalent models can be found in reference [1] The paper concludes with directions for future research.
A method of noise reduction based on discrimination function minimization (DFM) is presented. This method requires the definition of a function that discriminates between clean and noisy speech. A particular discrimin...
详细信息
A method of noise reduction based on discrimination function minimization (DFM) is presented. This method requires the definition of a function that discriminates between clean and noisy speech. A particular discrimination function is derived for a certain type of noise, such that the minimization of this function reduces the noise in the noisy speech signal. The DFM noise reduction method does not assume stationarity of the noise over an entire speech period. It is shown how an abrupt change in piecewise-stationary noise can be tracked. The method has been successfully applied to the reduction of automobile noise.< >
An approach is presented for efficiently designing and implementing FIR (finite impulse response) filter banks with equal passband bandwidths. The overall bank of L filters is generated from a prototype 2Lth-band filt...
详细信息
An approach is presented for efficiently designing and implementing FIR (finite impulse response) filter banks with equal passband bandwidths. The overall bank of L filters is generated from a prototype 2Lth-band filter with the aid of a few additional coefficients and adders. The attractive properties of these filter banks are that the sum of filter outputs is guaranteed to be a pure delay and the passband bandwidths are the same for all the filters.< >
The theory of mathematical morphology is presented. The author introduces the basic morphological transformations and reviews some of their properties. He also summarizes the theory of morphological filtering. Several...
详细信息
The theory of mathematical morphology is presented. The author introduces the basic morphological transformations and reviews some of their properties. He also summarizes the theory of morphological filtering. Several characteristics of mathematical morphology which demonstrate its applicability in the processing of medical images are outlined.< >
The authors outline a single (two-) microphone design of an ANC (adaptive noise cancellation) system that avoids prediction of colored noise, thus achieving the optimal prediction residual that is theoretically achiev...
详细信息
The authors outline a single (two-) microphone design of an ANC (adaptive noise cancellation) system that avoids prediction of colored noise, thus achieving the optimal prediction residual that is theoretically achievable. The design forces the colored acoustic noise to its inaccessible white generating sequence. The system, which is less sensitive to parameter estimation errors than previous approaches, is easily extended to an array of cancellation speakers. computer simulation results using actually recorded machine noise are presented to illustrate (simulated) system performance. Parameter sensitive analysis shows that the present design is far less sensitive to identification than are deterministic designs.< >
An algorithm for detecting neural processes in serial optical sections for use in an automated three-dimensional neural reconstruction system is presented. This parsimonious, nonlinear, psychophysically motivated algo...
详细信息
An algorithm for detecting neural processes in serial optical sections for use in an automated three-dimensional neural reconstruction system is presented. This parsimonious, nonlinear, psychophysically motivated algorithm addresses the problems specific to neural element detection and localization, viz., images with minimal resolution, operators with small spatial supports, highly curved, filamentous features, large variation in feature intensity profile, poor signal-to-noise ratio, and determination of depth without stereo. One first finds the magnitude and orientation of the maximum intensity second directional derivative. A family of curves is locally fitted to these data, and the projections of the data on the curve family are found. If a pixel lies on a curve with sufficient total projection, it is labeled with the magnitude, orientation, curvature, spatial extent, and element displacement. Depth is interpolated from the spatial extent data for corresponding neighborhoods in three adjacent (in depth) images by using an approximation to the depth-dependent optical point spread function. Experimental results using photomicrographs of cat visual cortex are presented.< >
A procedure for the reformulation of Chang's criterion (1981) for the absence of zero-input limit cycles in recursive digital filters is presented. The technique employs the standard bilinear transform to convert ...
详细信息
A procedure for the reformulation of Chang's criterion (1981) for the absence of zero-input limit cycles in recursive digital filters is presented. The technique employs the standard bilinear transform to convert a frequency-domain criterion into a polynomial form in terms of matrix transformations. The use of the method is illustrated with design examples applied to determine the stability region for certain single-quantizer recursive filter structures. The results can be utilized, e.g., in the design of an error feedback network for a given filter such that limit cycles are eliminated.< >
A multichannel Givens lattice algorithm is presented, which has a modular structure and is suitable for parallel implementation. The basic processing module requires a 2*2 plane rotation (Givens rotation), which has s...
详细信息
A multichannel Givens lattice algorithm is presented, which has a modular structure and is suitable for parallel implementation. The basic processing module requires a 2*2 plane rotation (Givens rotation), which has superior numerical properties due to its circular nature. The application of the multichannel Givens lattice algorithm to channel equalization is also discussed.< >
暂无评论