marytts.util.math

Class DTW

java.lang.Object

marytts.util.math.DTW

public class DTW
extends Object

Dynamic programming to compute similarity measure

Author:

sathish

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	DTW.Node
class	DTW.RecurssiveDTW

Constructor Summary

Constructors

Constructor and Description
DTW(double[][] signal, double[][] reference) Dynamic time warping (DTW) cost signal and reference Default 'Euclidean' distance function
DTW(double[][] signal, double[][] reference, double[] sigma2) DTW using Mahalanobis distance (Variance computation from external module)
DTW(double[][] signal, double[][] reference, String distanceFunction) Dynamic time warping (DTW) cost signal and reference distanceFunction = {"Euclidean" or "Absolute"}

Method Summary

Modifier and Type	Method and Description
double	AbsDistance(double[] x, double[] y) Absolute distance
double	EuclideanDistance(double[] x, double[] y) Euclidean distance
protected double	frameDistance(double[] f1, double[] f2, String distanceType)
double	getCost() Get cost of best path
double	getNormalizedCost() Get cost of best path
double[]	weightFunction(int windowLength)

Methods inherited from class java.lang.Object

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

Constructor Detail

DTW

public DTW(double[][] signal,
           double[][] reference)

Dynamic time warping (DTW) cost signal and reference Default 'Euclidean' distance function

Parameters:

signal - signal

reference - reference

DTW

public DTW(double[][] signal,
           double[][] reference,
           String distanceFunction)

Dynamic time warping (DTW) cost signal and reference distanceFunction = {"Euclidean" or "Absolute"}

Parameters:

signal - signal

reference - reference

distanceFunction - distance function

DTW

public DTW(double[][] signal,
           double[][] reference,
           double[] sigma2)

DTW using Mahalanobis distance (Variance computation from external module)

Parameters:

signal - signal

reference - reference

sigma2 - sigma2

Method Detail

getCost

public double getCost()

Get cost of best path

Returns:

cost

getNormalizedCost

public double getNormalizedCost()

Get cost of best path

Returns:

cost

EuclideanDistance

public double EuclideanDistance(double[] x,
                                double[] y)

Euclidean distance

Parameters:

x - x

y - y

Returns:

sum

AbsDistance

public double AbsDistance(double[] x,
                          double[] y)

Absolute distance

Parameters:

x - x

y - y

Returns:

sum

weightFunction

public double[] weightFunction(int windowLength)

frameDistance

protected double frameDistance(double[] f1,
                               double[] f2,
                               String distanceType)