Degradation of speech signal due to adverse conditions is the major challenge for automatic speech recognition (ASR) systems. This paper introduces a novel approach to estimate an Ideal Neighbourhood Mask (INM) for speech segregation based on deep neural networks estimator. The method described here is based on the local binary patterns (LBP) technique often used in digital image processing. Ideal Neighbourhood Mask will indicate which time-frequency (T-F) units of the noisy speech are canceled. The performance assessment of the proposed application in conjunction with the traditional mask techniques, i.e., Ideal Binary Mask (IBM) and Ideal Ratio Mask (IRM), are carried out under various environments regarding the objective speech quality measures. The recognition experiments including results in the AURORA IV framework indicate that the proposed scheme, when applied in adverse environments yield significantly better performance than the conventional techniques.

In this paper, a text-independent automatic speaker recognition (ASkR) system is proposed-the SR/sub Hurst/-which employs a new speech feature and a new classifier. The statistical feature pH is a vector of Hurst (H) parameters obtained by applying a wavelet-based multidimensional estimator (M/spl I.bar/dim/spl I.bar/wavelets ) to the windowed short-time segments of speech. The proposed classifier for the speaker identification and verification tasks is based on the multidimensional fBm (fractional Brownian motion) model, denoted by M/spl I.bar/dim/spl I.bar/fBm. For a given sequence of input speech features, the speaker model is obtained from the sequence of vectors of H parameters, means, and variances of these features. The performance of the SR/sub Hurst/ was compared to those achieved with the Gaussian mixture models (GMMs), autoregressive vector (AR), and Bhattacharyya distance (dB) classifiers. The speech database-recorded from fixed and cellular phone channels-was uttered by 75 different speakers. The results have shown the superior performance of the M/spl I.bar/dim/spl I.bar/fBm classifier and that the pH feature aggregates new information on the speaker identity. In addition, the proposed classifier employs a much simpler modeling structure as compared to the GMM.

In this paper, we introduce a formula to align the vertical coefficients of the SA-DCT (shape-adaptive discrete cosine transform). Instead of grouping coefficients with the same index as proposed by Sikora and Makai (1995), the new method employs an alignment by phase strategy. Experimental results are given for both synthetic segments and real testing image sequences.

This paper presents a mathematical formulation of shape-adaptive (SA) two-dimensional (2-D) transforms. A generic SA transform is represented by a concatenation of three linear operators that characterize the one-dimensional (1-D) vertical transforms, the alignment of vertical coefficients, and the 1-D horizontal transforms. We state and prove several powerful properties of these operators. The mathematical formulation described may serve as an important theoretical basis for future developments in the area of object-based video coding.

This article presents a new strategy to encode the LP (linear predictive) short-time spectral envelope (STSE) of speech. A better reconstruction of the STSE is achieved by modifying the usual trade-off between the transmission rate of the LP parameters and the performance of the quantization algorithm. A differential coding based on bidirectional prediction and hybrid vector quantization is used to compensate the increase in transmission rate. Simulation results show the effectiveness of this coding strategy.

We propose an adaptive sub-band image coder designed according to subjective criteria. A pre-processing technique is applied to the high subbands and discards local information irrelevant to the overall subjective quality. Pixels not removed by this technique are classified as edges and non-edges by means of a specific algorithm designed for each sub-band. The two classes in each high-frequency sub-band are quantized with entropy constrained vector quantizers (ECVQ). The low sub-band is encoded with an edge-preserving ADPCM. Both in terms of PSNR and visual quality the proposed codec offers meaningful improvements compared to the JPEG algorithm.

In this letter, we propose a CELP algorithm in which the complexity of the search procedure in the adaptive codebook is greatly reduced. This is achieved by means of a modified model of the CELP synthesizer, while keeping the usual perceptual weighting of the synthesis error in the analysis procedure. Simulation results show that the proposed algorithm can provide speech quality comparable to the one obtained with the conventional CELP codec.>

We propose a method to identify the dominant coefficients in DCT image coders at a very low rate. The method consists of applying a weighted vector quantization technique to a binary-valued matrix in which the 1's indicate the positions of the dominant DCT coefficients. In this technique the distortion measure is inversely proportional to a weighted number of coincident 1's, where larger weighting factors are assigned to coincident 1's in the low-frequency region. Simulation results show that only 0.16 bit/pixel is sufficient to identify the dominant DCT coefficients in a fixed rate scheme, without any significant degradation of the reconstructed image.>

Examines the application of simulated annealing to assign indices to the output levels of a vector quantizer (VQ). The quantizer is used for the LSF parameter representation of the short-term predictor of a CELP coder. Three classes of error protection and a 0.1 channel error rate are considered. Simulation results show that the use of the index assignment provides significant gains both in terms of average distortion as well as in terms of reduction of large outliers. The improvements afforded by the index assignment, however, were not significant when the VQ was used inside the structure of a CELP coder.>