Abstract
A discussion of linear prediction analysis methods that employ the Cholesky decomposition is given. Only one method insures stability of the synthesis filters by assuring that estimated reflection coefficients are bounded in magnitude by one. This estimate is generalized to one based on forward and backward prediction errors. An efficient computational algorithm based on Cholesky factorization is derived.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have