An improved high precision formant analysis procedure

Describes an improved method and it´s implementation for the extraction of formants from high order autoregressive models, using previously described techniques. Test results are presented for synthetic and real data and it is shown that success rates of over 80% are now obtainable for model orders as high as 200. The procedure described will find the factors of a polynomial using signal processing techniques rather than numerical methods, which are unreliable for such high order models. The algorithm accepts an LPC model of a sample of a drum beat, found by any conventional method, for analysis