org.xiph.speex
Class Lpc

java.lang.Object
  org.xiph.speex.Lpc

public class Lpc

extends java.lang.Object

LPC - and Reflection Coefficients

The next two functions calculate linear prediction coefficients and/or the related reflection coefficients from the first P_MAX+1 values of the autocorrelation function.

Invented by N. Levinson in 1947, modified by J. Durbin in 1959.

Version:

$Revision: 1.3 $

Author:

Marc Gimpel, Wimba S.A. (marc@wimba.com)

Constructor Summary

Lpc()

Method Summary

static void	autocorr(float[] x, float[] ac, int lag, int n) Compute the autocorrelation ,--, ac(i) = > x(n) * x(n-i) for all n `--' for lags between 0 and lag-1, and x == 0 outside 0...n-1
static float	wld(float[] lpc, float[] ac, float[] ref, int p) Returns minimum mean square error.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Lpc

public Lpc()

Method Detail

wld

public static float wld(float[] lpc,
                        float[] ac,
                        float[] ref,
                        int p)

Returns minimum mean square error.

Parameters:

lpc - - float[0...p-1] LPC coefficients

ac - - in: float[0...p] autocorrelation values

ref - - out: float[0...p-1] reflection coef's

p -

Returns:

minimum mean square error.

autocorr

public static void autocorr(float[] x,
                            float[] ac,
                            int lag,
                            int n)

Compute the autocorrelation ,--, ac(i) = > x(n) * x(n-i) for all n `--' for lags between 0 and lag-1, and x == 0 outside 0...n-1

Parameters:

x - - in: float[0...n-1] samples x

ac - - out: float[0...lag-1] ac values

lag -

n -