A Model of LPC Excitation in Terms of Eigenvectors of the Autocorrelation Matrix of the Impulse Response of the LPC Filter

Linear predictive coding methods require efficient representation of both the LPC filter and its input excitation of synthesize high quality speech at low bit rates. Although important progress has been made in encoding the LPC filter parameters, it is still not possible to encode the filter excitation at low bit rates and still maintain high voice quality in the synthetic speech signal. As a first step in achieving low bit rates, this paper describes a framework for developing a parametric representation that can provide a compact yet accurate representation of the excitation.