Physical significance of least mean squares criterion in nonstationary lattice predictors
Abstract
An acoustical nonstationary model of the vocal tract is discussed. The effectiveness of the nonstationary lattice predictor in recovering the model is illustrated. The reflection coefficients are linearly decomposed on a function basis and the weights of the linear combinations are the new parameters of the model. Parameters are determined by an algorithm similar to Burg's by minimizing the sum of the variances of the forward and backward residuals. Synthetic signals of good approximation are obtained in two examples.
 Publication:

NASA STI/Recon Technical Report N
 Pub Date:
 June 1985
 Bibcode:
 1985STIN...8631789M
 Keywords:

 Approximation;
 Least Squares Method;
 Mathematical Models;
 Voice Communication;
 Voice Data Processing;
 Algorithms;
 Lattices (Mathematics);
 Predictions;
 Systems Stability;
 Vocal Cords;
 Communications and Radar