edu.cmu.sphinx.linguist.acoustic.tiedstate

Class MixtureComponent

java.lang.Object

edu.cmu.sphinx.linguist.acoustic.tiedstate.MixtureComponent

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable

Direct Known Subclasses:

PrunableMixtureComponent

public class MixtureComponent
extends java.lang.Object
implements java.lang.Cloneable, java.io.Serializable

Defines the set of shared elements for a GaussianMixture. Since these elements are potentially shared by a number of GaussianMixtures, these elements should not be written to. The GaussianMixture defines a single probability density function along with a set of adaptation parameters.

Note that all scores and weights are in LogMath log base

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static float	DEFAULT_DIST_FLOOR
static float	DEFAULT_VAR_FLOOR
protected float	distFloor
protected float	logPreComputedGaussianFactor
protected float[]	meanTransformed Mean after transformed by the adaptation parameters.
protected float[]	precisionTransformed Precision is the inverse of the variance.

Constructor Summary

Constructors

Constructor and Description
MixtureComponent(float[] mean, float[] variance) Create a MixtureComponent with the given sub components.
MixtureComponent(float[] mean, float[][] meanTransformationMatrix, float[] meanTransformationVector, float[] variance, float[][] varianceTransformationMatrix, float[] varianceTransformationVector) Create a MixtureComponent with the given sub components.
MixtureComponent(float[] mean, float[][] meanTransformationMatrix, float[] meanTransformationVector, float[] variance, float[][] varianceTransformationMatrix, float[] varianceTransformationVector, float distFloor, float varianceFloor) Create a MixtureComponent with the given sub components.

Method Summary

Modifier and Type	Method and Description
MixtureComponent	clone()
float[]	getMean() Returns the mean for this component.
float	getScore(float[] feature) Calculate the score for this mixture against the given feature.
float	getScore(FloatData feature) Calculate the score for this mixture against the given feature.
float[]	getVariance() Returns the variance for this component.
float	precomputeDistance() Pre-compute factors for the Mahalanobis distance.
java.lang.String	toString()
void	transformStats() Applies transformations to means and variances.

Methods inherited from class java.lang.Object

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

Field Detail

meanTransformed

protected float[] meanTransformed

Mean after transformed by the adaptation parameters.

precisionTransformed

protected float[] precisionTransformed

Precision is the inverse of the variance. This includes adaptation.

distFloor

protected float distFloor

DEFAULT_VAR_FLOOR

public static final float DEFAULT_VAR_FLOOR

See Also:

Constant Field Values

DEFAULT_DIST_FLOOR

public static final float DEFAULT_DIST_FLOOR

See Also:

Constant Field Values

logPreComputedGaussianFactor

protected float logPreComputedGaussianFactor

Constructor Detail

MixtureComponent

public MixtureComponent(float[] mean,
                        float[] variance)

Create a MixtureComponent with the given sub components.

Parameters:

mean - the mean vector for this PDF

variance - the variance for this PDF

MixtureComponent

public MixtureComponent(float[] mean,
                        float[][] meanTransformationMatrix,
                        float[] meanTransformationVector,
                        float[] variance,
                        float[][] varianceTransformationMatrix,
                        float[] varianceTransformationVector)

Create a MixtureComponent with the given sub components.

Parameters:

mean - the mean vector for this PDF

meanTransformationMatrix - transformation matrix for this pdf

meanTransformationVector - transform vector for this PDF

variance - the variance for this PDF

varianceTransformationMatrix - var. transform matrix for this PDF

varianceTransformationVector - var. transform vector for this PDF

MixtureComponent

public MixtureComponent(float[] mean,
                        float[][] meanTransformationMatrix,
                        float[] meanTransformationVector,
                        float[] variance,
                        float[][] varianceTransformationMatrix,
                        float[] varianceTransformationVector,
                        float distFloor,
                        float varianceFloor)

Create a MixtureComponent with the given sub components.

Parameters:

mean - the mean vector for this PDF

meanTransformationMatrix - transformation matrix for this pdf

meanTransformationVector - transform vector for this PDF

variance - the variance for this PDF

varianceTransformationMatrix - var. transform matrix for this PDF

varianceTransformationVector - var. transform vector for this PDF

distFloor - the lowest score value (in linear domain)

varianceFloor - the lowest value for the variance

Method Detail

getMean

public float[] getMean()

Returns the mean for this component.

Returns:

the mean

getVariance

public float[] getVariance()

Returns the variance for this component.

Returns:

the variance

getScore

public float getScore(FloatData feature)

Calculate the score for this mixture against the given feature.

Note: The support of DoubleData-features would require an array conversion to float[]. Because getScore might be invoked with very high frequency, features are restricted to be FloatDatas.

Parameters:

feature - the feature to score

Returns:

the score, in log, for the given feature

getScore

public float getScore(float[] feature)

Calculate the score for this mixture against the given feature. We model the output distributions using a mixture of Gaussians, therefore the current implementation is simply the computation of a multi-dimensional Gaussian.

Normal(x) = exp{-0.5 * (x-m)' * inv(Var) * (x-m)} / {sqrt((2 * PI) ^ N) * det(Var))}

where x and m are the incoming cepstra and mean vector respectively, Var is the Covariance matrix, det() is the determinant of a matrix, inv() is its inverse, exp is the exponential operator, x' is the transposed vector of x and N is the dimension of the vectors x and m.

Parameters:

feature - the feature to score

Returns:

the score, in log, for the given feature

precomputeDistance

public float precomputeDistance()

Pre-compute factors for the Mahalanobis distance. Some of the Mahalanobis distance computation can be carried out in advance. Specifically, the factor containing only variance in the Gaussian can be computed in advance, keeping in mind that the the determinant of the covariance matrix, for the degenerate case of a mixture with independent components - only the diagonal elements are non-zero - is simply the product of the diagonal elements.

We're computing the expression:

{sqrt((2 * PI) ^ N) * det(Var))}

Returns:

the precomputed distance

transformStats

public void transformStats()

Applies transformations to means and variances.

clone

public MixtureComponent clone()
                       throws java.lang.CloneNotSupportedException

Overrides:

clone in class java.lang.Object

Throws:

java.lang.CloneNotSupportedException

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object