Informatica

The paper presents a novel algorithm for restoration of the missing samples in additive Gaussian noise based on the forward–backward autoregressive (AR) parameter estimation approach and the extrapolation technique. The proposed algorithm is implemented in two consecutive steps. In the first step, the forward–backward approach is used to estimate the parameters of the given neighbouring segments, while in the second step the extrapolation technique for the segments is applied to restore the samples of the missing segment. The experimental results demonstrate that the restoration error of the samples of the missing segment using the proposed algorithm is reduced as compared with the Burg algorithm.

The paper presents a novel method for improving the estimates of closely-spaced frequencies of a short length signal in additive Gaussian noise based on the Burg algorithm with extrapolation. The proposed method is implemented in two consecutive steps. In the first step, the Burg algorithm is used to estimate the parameters of the predictive filter, while in the second step the extrapolation technique of the signal is used to improve the frequency estimates. The experimental results demonstrate that the frequency estimates of the short length signal, using the Burg algorithm with extrapolation, are more accurate than the frequency estimates using the Burg algorithm without extrapolation.

The paper addresses the problem of discrimination of homographs when a lengthy segment of an uttered word is missing. The considered discrimination procedure is done by recognizer that operates on cepstrum coefficients extracted from the speech signal. For restoration of the missing speech segment rather than use of the known speech signal, it has been proposed to calculate speech signal characteristics: the period of fundamental frequency and intensity. By experimentation it has been shown that the polynomial approximation of speech signal characteristics improves homograph discrimination results. An extra computational burden associated with the proposed method is not high because it involves recalculation of the already extracted Fourier coefficients.

In this paper on basis of the results (Dyomin et al., 2003a) the structure of Shannon information amount in the joint filtering and extrapolation problem of the stochastic processes by continuous‐discrete time memory observations is investigated. For particular class of processes with applying of the general results the problem of optimal transmission over the lag channels is considered and efficiency of filtering and extrapolation receptions under transmission over channels with memory or lag is investigated.

This paper considers an information aspect of the problem of the joint filtering and generalized extrapolation, when the output of observation channels (data transmission) is the realizations set of the processes with continuous and discrete time, which depend on both the current and the past values of unobservable process (useful signal). The relations defining time evolution of Shannon information are obtained. The particular problems of the memory channels information efficiency and optimal transmission of stochastic processes, with applying the general results are considered.