Main Content

# LSF/LSP to LPC Conversion

Convert line spectral frequencies or line spectral pairs to linear prediction coefficients

Libraries:
DSP System Toolbox / Estimation / Linear Prediction

## Description

The LSF/LSP to LPC Conversion block converts line spectral pairs (LSPs) or line spectral frequencies (LSFs) to linear prediction polynomial coefficients (LPCs). When converting LSFs to LPCs, the block outputs match those of the lsf2poly function.

## Ports

### Input

expand all

Specify the line spectral pairs as a vector of values in the range (−1 1) in decreasing order. If you specify a matrix, the block treats each column of the matrix as a separate channel. When you specify an unoriented vector, the block treats the input as one channel. These values equal the cosine of the LSF values between 0 and π radians. Do not include the guaranteed LSP values, −1 and 1.

Data Types: single | double

Specify the line spectral frequencies normalized in the range (0 0.5) as a vector or a matrix in increasing order. If you specify a matrix, the block treats each column of the matrix as a separate channel. When you specify an unoriented vector, the block treats the input as one channel. You can obtain these values by dividing the LSF values between 0 and π radians by 2π. Do not include the guaranteed LSF values, 0 and 0.5.

Data Types: single | double

Specify the line spectral frequencies in the range (0 π) as a vector or a matrix in increasing order. If you specify a matrix, the block treats each column of the matrix as a separate channel. When you specify an unoriented vector, the block treats the input as one channel. Do not include the guaranteed LSF values, 0 and π.

Data Types: single | double

### Output

expand all

Linear prediction coefficients, returned as a vector or a matrix.

Data Types: single | double

## Parameters

expand all

When you select this parameter to:

• LSP in range (-1 1) –– The block converts the line spectral pairs in the range (−1 1) to linear prediction coefficients.

• LSF in range (0 pi) –– The block converts the line spectral frequencies in the range (0 π) to linear prediction coefficients.

• LSF normalized in range (0 0.5) –– The block converts the line spectral frequencies normalized in the range (0 0.5) to linear prediction coefficients.

## Block Characteristics

 Data Types double | single Direct Feedthrough no Multidimensional Signals no Variable-Size Signals no Zero-Crossing Detection no

## References

[1] Kabal, P. and Ramachandran, R. “The Computation of Line Spectral Frequencies Using Chebyshev Polynomials.” IEEE Transactions on Acoustics, Speech, and Signal Processing, Vol. ASSP-34 No. 6, December 1986. pp. 1419-1426.

## Version History

Introduced before R2006a