Title :
Algorithm for high precision root finding from high order LPC models
Author_Institution :
Dept. of Electron. & Electr. Eng., King´´s Coll., London, UK
Abstract :
A method for the extraction of formants from high order autoregressive models and its implementation are described. Test results are presented for synthetic and real data. It is shown that success rates of the order of 70% are obtainable even for model orders as high as 200. The algorithm described has been developed for the particular purpose of finding the resonant frequencies (formants) in the acoustic waveforms of atonal percussion instruments (drums). The basic physical model has much in common with that used for speech coding and analysis, but there are significant differences in the two types of waveforms. This leads to slightly different applications of the fundamental theory, implemented using the linear predictive coding (LPC) model, as applied to drum samples.<>
Keywords :
acoustic signal processing; encoding; filtering and prediction theory; musical instruments; acoustic waveforms; algorithm; atonal percussion instruments; drum samples; drums; formants extraction; high order LPC models; high order autoregressive models; high precision root finding; linear predictive coding; real data; resonant frequencies; speech analysis; speech coding; success rates; synthetic data;
Journal_Title :
Communications, Speech and Vision, IEE Proceedings I
