The computation of line spectral frequency using the second Chebyshev polynomials

Recently, there has been a growing interest in the use of line spectral frequency (LSF) to code the filter parameters for linear predictive coding (LPC) of speech. We present a new version of the algorithm used by J. Rothweiler for computing LSF coefficients (see IEEE Trans. Speech and Audio Process...

Full description

Saved in:
Bibliographic Details
Main Authors: Yao Tianren, Xiang Juanjuan, Lu Wei
Format: Conference Proceeding
Language:English
Subjects:
Chebyshev approximation
Frequency
Linear predictive coding
Nonlinear filters
Polynomials
Quantization
Reflection
Spectral shape
Speech analysis
Speech coding
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Recently, there has been a growing interest in the use of line spectral frequency (LSF) to code the filter parameters for linear predictive coding (LPC) of speech. We present a new version of the algorithm used by J. Rothweiler for computing LSF coefficients (see IEEE Trans. Speech and Audio Processing, vol.7, no.5, 1999). This new version uses the second Chebyshev polynomials to reduce the function order. Theoretical research indicates that the improved algorithm can lead to more concise expressions, eliminate multiplication and reduce the complexity of computation further when the linear prediction order increases.
DOI:10.1109/ICOSP.2002.1181023