Title :
A novel algorithm to estimate the line spectral frequencies from LPC coefficients
Author :
Nakhai, Mohammad R. ; Marvasti, Farokh A.
Author_Institution :
Dept. of Electron. Eng., London Univ., UK
fDate :
31 May-3 Jun 1998
Abstract :
A new method to estimate the line spectral frequencies (LSF´s) with any desired accuracy is described. LSF´s are estimated by locating the zeros of two polynomials derived from LPC inverse filter polynomial on the unit circle in z-plane. The computational load increases as the required accuracy increases. An important feature of algorithm is the ability of compromising the required accuracy and computational load. The method can also be applied to convert the higher order LPC coefficients to associated LSF´s. The algorithm does not require a very high precision arithmetic and can be implemented using fixed-point digital signal processors
Keywords :
estimation theory; poles and zeros; polynomials; signal processing; LPC coefficients; LPC inverse filter polynomial; computational load; fixed-point digital signal processors; line spectral frequencies; unit circle; z-plane; Dynamic range; Filters; Frequency estimation; Linear predictive coding; Polynomials; Quantization; Signal processing algorithms; Speech analysis; Speech coding; Telecommunication computing;
Conference_Titel :
Circuits and Systems, 1998. ISCAS '98. Proceedings of the 1998 IEEE International Symposium on
Conference_Location :
Monterey, CA
Print_ISBN :
0-7803-4455-3
DOI :
10.1109/ISCAS.1998.698793