In this paper we present an approach for incorporating a state of the art speech recognition engine into a novel assistive reading system for Greek dyslexic students. This system is being developed in the framework of the AGENT-DYSL IST project, and facilitates dyslexic children in learning to read fluently. Unlike previously presented approaches, the aim of this system is to monitor the progress and perspectives of a dyslexic user and supply personalised help. The goal of this help is to gradually increase the reading capabilities of the user, gradually diminish the assistance provided, till he is able to read as a non-dyslexic reader. Copyright © 2011 ISCA.

In this paper, a Decimative Spectral estimation method based on Eigenanalysis and SVD (Singular Value Decomposition) is presented and applied to speech signals in order to estimate Formant/Bandwidth values. The underlying model decomposes a signal into complex damped sinusoids. The algorithm is applied not only on speech samples but on a small amount of the autocorrelation coefficients of a speech frame as well, for finer estimation. Correct estimation of Formant/Bandwidth values depend on the model order thus, the requested number of poles. Overall, experimentation results indicate that the proposed methodology successfully estimates formant trajectories and their respective bandwidths.

A new state-space method for spectral estimation that performs decimation by factor two while it makes use of the full set of data available is presented. The proposed method, called DESE2, is based on singular value decomposition in order to estimate frequency, damping factor, amplitude and phase of exponentially damped sinusoids in the presence of noise. The DESE2 method is compared against some previously proposed methods for spectral estimation that lie among the most promising methods in the field of spectroscopy, where accuracy of parameter estimation is of utmost importance. Experiments performed on a typical simulated NMR signal prove the new method to be more robust, especially for low signal to noise ratio. The new method outperforms the other two not only by presenting lower failure rates but also by incorporating enhanced discriminative analysis while at the same time it benefits from the use of the full data set.

This paper presents a new state-space method for spectral estimation based on a companion matrix technique in order to estimate frequency, damping factor, amplitude and phase of exponential sinusoids. The new method, called CSE (companion matrix based spectral estimation), is compared against a previously proposed method called HTLS which is based on the use of total least squares. The latter lies among the most promising methods in the field of spectroscopy where accuracy of parameter estimation is of utmost importance. Experiments performed on a simulated NMR signal prove CSE to be more robust in terms of failure rate, especially for low signal to noise ratio.

Harmonic selection is a useful mechanism for speech enhancement. However, there is difficulty in detecting harmonic structure in the spectrum in the presence of noise or interfering speech. In this paper we address the problem of applying the reassignment method to produce a higher-resolution spectrum for improving the harmonic selection. The reassignment method is to assign the value of the spectrogram computed by the short-time Fourier transform to the gravity center of the region rather than the geometric center of the region. We analyze the resolution capability of the reassigned Fourier spectrum and show that it has better frequency separation than Fourier spectrum. We incorporate the reassignment method with the amplitude-modulation based harmonic selection to segregate a target vowel from interfering vowels. Experimental results show that the target vowel recognition performance is improved by using the reassignment method for increasing the readability of the harmonic structure in speech signals.

This paper presents two different directions to build HMM models which give enough acoustic resolution and t in limited user resources. They both refer to scaling down the acoustic models which are built with tied gaussian HMMs. The total number of gaussians is reduced by a pairwise merging, and the number of gaussians per state is reduced by selecting them based on the so called occupancy criterion. Experiments carried out on the WSJ recognition task show that after scaling down, no further training is needed when the number of gaussians or the number of gaussians per state is reduced up to a factor three. This is an advantage as retraining can not be executed by the final system user.

We present a new speech coding algorithm, based on an all-pole model of the vocal tract. Whereas current autoregressive (AR) based modeling techniques (e.g. CELP, LPC-10) minimize a prediction error, which is considered to be the input to the all-pole model, our approach determines the closest (in L/sub 2/ norm) signal, which exactly satisfies an all-pole model. Each frame is then encoded by storing the parameters of the complex damped exponentials deduced from the all-pole model and its initial conditions. Decoding is performed by adding the complex damped exponentials based on the transmitted parameters. The new algorithm is demonstrated on a speech signal. The quality is compared with that of a standard coding algorithm at comparable compression ratios, by using the segmental signal-to-noise ratio (SNR).

This paper proposes a method to transform acoustic models that have been trained with a certain group of speakers for use on different speech in hidden Markov model based (HMM-based) automatic speech recognition. Features are transformed on the basis of assumptions regarding the difference in vocal tract length between the groups of speakers. First, the vocal tract length (VTL) of these groups has been estimated based on the average third formant F/sub 3/. Second, the linear acoustic theory of speech production has been applied to warp the spectral characteristics of the existing models so as to match the incoming speech. The mapping is composed of subsequent nonlinear submappings. By locally linearizing it and comparing results in the output, a linear approximation for the exact mapping was obtained which is accurate as long as the warping is reasonably small. The feature vector, which is computed from a speech frame, consists of the mel scale cepstral coefficients (MFCC) along with delta and delta/sup 2/-cepstra as well as delta and delta/sup 2/ energy. The method has been tested for TI digits data base, containing adult and children speech, consisting of isolated digits and digit strings of different length. The word error rate when trained on adults and tested on children with transformed adult models is decreased by more than a factor of two compared to the nontransformed case.

This paper proposes a method to transform acoustic models (HMM gaussian mixtures) that have been trained on a certain group of speakers for use on speech from a different group of speakers. Cepstral features are transformed on the basis of assumptions regarding the difference in vocal tract length (VTL) between the groups of speakers (VTL normalisation, VTLN). Firstly, the VTL of these groups has been estimated based on the average third formant F . Secondly, the linear acoustic theory of speech production has been applied to warp the spectral characteristics of the existing models so as to match the incoming speech. The mapping is composed of subsequent non-linear submappings. By locally linearizing it, a linear approximation was obtained which is accurate as long as warping is reasonably small. The method has been tested for the TI digits database, containing adult and kids speech, consisting of isolated digits and digit strings of different length. The word error rate when trained on adults and tested on kids with transformed adult models is decreased by more than a factor of 2 compared to the non-transformed case.

We show how successive projection-like algorithms may be used for approximation or exact modelling of a signal. For this purpose, we propose a new efficient algorithm providing adequate linear difference equations satisfied by the original signal. The projection operators at each step of the approximation algorithm and the new procedure are shown to be orthogonal. Finally, a pyramidal structure summarises the possibilities offered by the combinations of both procedures.

The paper describes the permutative vector quantization (PVQ) scheme as a special case of a more general structurally constrained vector quantization concept. This concept makes it possible to increase the vector dimensions beyond the technical bounds of conventional VQ and to exploit, by means of this, the inter-pixel correlations in large image blocks. Furthermore, a codebook design algorithm adapted to permutative VQ is proposed and it is shown experimentally that the coding performance of conventional VQ can be improved using the present scheme.

The frequency analysis of a signal is considered. Although the classical approach based on the representation of a signal in terms of complex exponentials has its merit, an alternative approach aiming to represent a signal by a sum of exponential functions is also interesting. An exact solution to this problem is found at the meeting point of linear predictive coding (LPC) and modeling techniques based on the singular value decomposition (SVD) analysis. Further insight in the structural analysis of a signal is provided. The rank of a signal is defined, and it is shown that full recovery of finite rank data from an incomplete set is always possible.>

This paper reports on a method for speech recognition using as pre-processing system either a Peripheral Auditory System (PAS) model, or Linear Prediction Coding (LPC). Certain speech recognition problems have been examined using Ficher's Discriminant analysis technic, in order to separate different classes of sounds.

This paper introduces a sequential fast scheme of the Kalman type to solve the covariance equations of linear prediction. The new algorithm constitutes a generalization of the scheme introduced recently by D. Falconer and L. Ljung, and applies to a more general situation. The algo rithm is characterized by conceptual simplicity and efficiency. Although it is based on matrix partitioning concepts and shifting properties of the signal vector, it does not have the restriction of prewindowing as it is the case with the previously mentioned algorithm, thus it can be used to model small frames of signals as well as to model signals generated by an AR model, giving in this case the exact solution. Finally it must be noted that the algorithm is strictly time recursive and leads to O(13p) multiplications per recursion only. Thus, this new scheme is the fastest existing method for the sequential solution of the general covariance equations when prewindowing is not desirable.