By Topic

Speech, Image Processing and Neural Networks, 1994. Proceedings, ISSIPNN '94., 1994 International Symposium on

Date 13-16 April 1994

Filter Results

Displaying Results 1 - 25 of 202
  • Proceedings of ICSIPNN '94. International Conference on Speech, Image Processing and Neural Networks

    Publication Year: 1994
    Request permission for commercial reuse | PDF file iconPDF (151 KB)
    Freely Available from IEEE
  • A robust and fast DP-CELP (double-pulse CELP) vocoder at the bit rate of 4 kb/s

    Publication Year: 1994, Page(s):563 - 566 vol.2
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (208 KB)

    It has been reported that in a CELP vocoder the perceptual quality of synthesized speech can be improved by replacing the traditional stochastic codebook with a mixed codebook containing periodic SP (single-pulse) and noise-like sequences. At the bit rate of 4 kb/s, the frame and subframe sizes are relatively long to reduce the bit rates for coding the gains and the synthesis filter parameters. In... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Novel quantization schemes for multi-pulse coder at 5 kbps

    Publication Year: 1994, Page(s):567 - 570 vol.2
    Cited by:  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (236 KB)

    A multi-pulse excitation model has been shown to produce very good quality speech at bit rates of 9.6 to 13 kbit/s. This model is seldom used in bit rates below 8 kbit/s because it requires many bits to quantize the pulse positions and amplitudes. The authors present several efficient quantization schemes to reduce the bit rate of a multi-pulse excitation coder. Subjective listening tests show tha... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Articulatory control of a vocal tract model based on fractional delay waveguide filters

    Publication Year: 1994, Page(s):571 - 574 vol.2
    Cited by:  Papers (2)  |  Patents (5)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (280 KB)

    A novel technique to implement and control an acoustic tube model of the human vocal tract is introduced. This model is an extension to the traditional Kelly-Lochbaum model, since not only the diameter of individual uniform tube sections but also their length, i.e., the positions of scattering junctions, can be continuously varied. The vocal tract model is implemented by means of FIR-type interpol... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Edge detection on real time using LOG filter

    Publication Year: 1994, Page(s):37 - 40 vol.1
    Cited by:  Papers (3)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (168 KB)

    A new technique is proposed for LOG (Laplacian Of Gaussian) filtering based on the method using look-up-tables. The authors take advantage of the circularly symmetric characteristic of the LOG filter, then a look-up-table with acceptable size is established. This transforms a large amount of multiplying operations into easy look-up-table operations. At the same time, another mask which can form an... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • CNN matching problem and applications

    Publication Year: 1994, Page(s):531 - 534 vol.2
    Cited by:  Papers (3)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (168 KB)

    Details of cellular neural network (CNN) universal machine and supercomputer have been presented (IEEE Trans. Circuits Syst.-I, vol.40, 1993). When cells in CNN are equipped with local logical memory or local analog memory, CNN can be used for solving matching problems. Matching with a supercomputer speed, CNN provides a rapid approach to search problems. CNN matching can also be used in real-time... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A hybrid neural network/rule based architecture for diphone speech synthesis

    Publication Year: 1994, Page(s):323 - 326 vol.1
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (228 KB)

    Analogue neural networks (ANNs) have successfully been applied to controlling a formant speech synthesiser, resulting in high quality speech. However they are somewhat limited by the large number of hidden layer neurons needed. The paper describes the application of a hybrid ANN/rule-based optimised computing architecture to diphone speech synthesis. The architecture utilises a simplified rule-bas... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Unstructured to structured error correction using neural nets

    Publication Year: 1994, Page(s):457 - 459 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (132 KB)

    Most transmitted or stored information are subjected to occasional errors. In most situations, the source of this information has inherent unstructured redundancy that can be exploited to correct these errors. In addition to the storage requirements, getting the source statistics required to perform the error correction may not be easy. In this paper, we propose and evaluate trained neural nets to... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Shape characterization and its applications

    Publication Year: 1994, Page(s):9 - 12 vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (280 KB)

    A new concept called the shape spectrum is proposed in this paper for shape characterization of image features. A multi-tolerance region growing method and a shape feature set are first introduced. Some general shape characterization methods, such as shape-factor-converted images, shape histogram, and shape spectrum, are then described. Preliminary results of application to synthesized images are ... View full abstract»

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

    Publication Year: 1994, Page(s):225 - 228 vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (216 KB)

    We are usually interested in training a connectionist model to converge either to a stable or an oscillatory state. Chaotic behaviour in connectionist models is rarely studied although evidences of chaos have already he found in many real life phenomena. In this work, we study the chaotic behaviour in recurrent networks by training a ring-structure recurrent model to learn a classical chaotic seri... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The fast algorithm for the finite length discrete wavelet transform

    Publication Year: 1994, Page(s):642 - 645 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (196 KB)

    The paper presents a structured algorithm for the finite length discrete wavelet transform. The analysis and synthesis filter matrices H, G can be decomposed in kronecker product form with cyclic block matrix and lower-triangle block matrix. The cyclic matrix can be implemented using FFT and the lower-triangle matrix is implemented straightforward. The arithmetic complexity of the algorithm is pri... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A comparison of acoustic and magnetic resonance imaging techniques in the estimation of vocal tract area functions

    Publication Year: 1994, Page(s):575 - 578 vol.2
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (212 KB)

    In the field of articulatory synthesis, the task of acoustic to geometric mapping requires a large number of vocal tract area functions. Magnetic resonance imaging has been used to determine such area functions, but is costly and time consuming. For this reason, an acoustic technique has been investigated to determine vocal tract shapes. In this method, the reflected wave resulting from an impulse... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Precision tracking of overlapping small targets

    Publication Year: 1994, Page(s):41 - 44 vol.1
    Cited by:  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (224 KB)

    The paper presents a Kalman filter based centroid tracking approach for multiple small targets. Before targets overlap, each target is tracked by a Kalman filter. When two targets overlap, a merged target is formed, and a Kalman filter for the overlapping target can be used to derive position and velocity of individual target. The main contributions of the paper are: 1) adaptive estimate of noise ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Signal analysis of electromyogram by artificial neural network

    Publication Year: 1994, Page(s):535 - 538 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (252 KB)

    During strong contraction, electromyogram (EMG) becomes a noise-like “interference pattern” composed of trains of motor-unit action potential (MUAP). With its adaptive properties, an artificial neural network (ANN) system is proposed and applied to the analysis of EMG for MUAP's detection. Features of MUAPs are extracted and fed into the ANN system for on-line training in which the num... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An adaptive pattern based image coding technique using visual pattern and texture pattern

    Publication Year: 1994, Page(s):180 - 183 vol.1
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (216 KB)

    VPIC (visual pattern image coding) and AMBTC (absolute moment block truncation coding) with LUT (look up table) are integrated to form a simple and fast system which yields images with very good visual quality while at moderately high compression rate. The VPs (visual patterns) are designed for subimages with prominent edges, while TPs (texture patterns) used in AMBTC are for high variance blocks.... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural network application for the analysis and synthesis of multilingual speech

    Publication Year: 1994, Page(s):327 - 330 vol.1
    Cited by:  Patents (3)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (212 KB)

    Presents a novel method for the extraction of parameters required to produce natural sounding speech. It has been shown to work for both French and English. The method adopted follows resonances within speech, rather than formants, furthermore resonances are allowed to cross under certain conditions. These parameters are then fed into a neural network based concatenation system allowing for the re... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Computer graphics animations of talking faces based on stochastic models

    Publication Year: 1994, Page(s):73 - 76 vol.1
    Cited by:  Papers (4)  |  Patents (8)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (268 KB)

    Animated computer graphics displays of the visible speech gestures of the human face have a number of potential applications. The paper describes a novel method for their creation by bringing together two statistically-based techniques, namely hidden Markov modelling and principal component analysis. The animations are derived from images of a real speaker's face and incorporate all the visible fe... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Handwritten character recognition by extended loop neural networks

    Publication Year: 1994, Page(s):460 - 463 vol.2
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (200 KB)

    Presents an extended loop neural network approach to handwritten character recognition. Experiments show that this method is very effective. The recognition rate by this method is higher than that by a backpropagation network View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Self-organization grouping for feature extraction and image segmentation

    Publication Year: 1994, Page(s):13 - 16 vol.1
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (296 KB)

    Feature extraction and image segmentation (FEIS) are two first goals of almost all image understanding systems. We think of FEIS as a multi-level process of recurrently grouping and describing at each abstraction level. We emphasize the role of grouping during this process because we believe that many features and events in real images are only perceived owing to the combination of weak evidence o... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Restoration of motion blurred images by MDL and piecewise constant image model

    Publication Year: 1994, Page(s):272 - 275 vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (192 KB)

    A novel algebraic approach to restore images blurred by motion is presented. Rather than using the LS or LMS criteria, the objective function is established on the basis of the Minimal Description Length (MDL) criterion and a piecewise constant model of underlying image. In this way, the a priori knowledge of the image can be included in the objective function. Computer simulations show that the m... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Transient responses in dynamical neural models

    Publication Year: 1994, Page(s):229 - 232 vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (180 KB)

    The authors consider the input/output behavior of a realistic dynamical neural model in comparison to those typically used in artificial neural networks. They have found that such models duplicate well those behaviors seen in living neurons, displaying a range of behaviors commonly seen in a wide variety of nonlinear dynamical systems. This is not captured well by weighted sum/monotonic transfer f... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Gabriel networks: self-organizing neural networks for adaptive vector quantization

    Publication Year: 1994, Page(s):658 - 661 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (220 KB)

    A self-organizing neural network model that resembles Kohonen's feature map model is presented in this paper. Unlike conventional feature maps which require static neighborhood relations to be defined a priori, our model is characterized by its use of dynamic neighborhood relations which change as learning proceeds. In particular, the neighborhood relations between neurons in a feature map are det... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A solution to the generalized Duda and Hart problem using Fourier parameterization

    Publication Year: 1994, Page(s):441 - 444 vol.2
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (188 KB)

    In a classic paper, R.O. Duda and P.E. Hart (1972) solved the unboundedness and non-uniformity problems of the line Hough transform by introducing a modified ρ-θ parameterization. Unfortunately, no equivalent parameterization exists for curves. W. Lam et al. (1993) proposed the Fourier descriptor as the parameterized curve equation in a Hough transform. We show that all parameters in thi... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Advances in image/video signal processing

    Publication Year: 1994, Page(s):VIII - VIXI vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (248 KB)

    The video signal is a sequence of images that requires real-time operation and a great deal of bandwidth. Video signal processing deals with filtering, matrixing, conversion, composition, bridging, and most importantly, compression. Various approaches for real-time video signal processing are presented. The pros and cons together with recent advances for each approach are also discussed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Parameter estimation of a fractional Brownian motion in a white noise using wavelets

    Publication Year: 1994, Page(s):646 - 649 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (224 KB)

    To discriminate the fractal parameter of a fractional Brownian motion (fBm) embedded in a white noise is equivalent to discriminating the composite singularity formed by superimposing a peak singularity upon a Dirac singularity. We use the autocorrelation of the wavelet transform coefficients to characterize the composite singularity, by formalizing this problem as a nonlinear optimization problem... View full abstract»

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