marytts.features

Class FeatureVector

java.lang.Object

marytts.features.FeatureVector

public class FeatureVector
extends Object

Compact representation of a list of byte-valued, short-valued and float-valued (continuous) features. In the user interface, features are identified through one index number, covering all three types of features. For example, a feature vector consisting of three bytes, four shorts and two floats will have nine features. Use getByteFeature(), getShortFeature() and getContinuousFeature() to access the features as primitives; note that you have to respect the index restrictions, i.e. in the example, calling getShortFeature(6) would be OK, but getShortFeature(2) or getShortFeature(8) would throw an IndexOutOfBoundsException. Use isShortFeature(i) to test whether a given feature is a short feature. Alternatively, you can use an Object interface to access all features in a uniform way: getFeature(i) will return a Number object for all valid indexes.

Author:

Marc Schröder

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	FeatureVector.FeatureType

Field Summary

Fields

Modifier and Type	Field and Description
byte[]	byteValuedDiscreteFeatures
float[]	continuousFeatures
short[]	shortValuedDiscreteFeatures
int	unitIndex

Constructor Summary

Constructors

Constructor and Description
FeatureVector(byte[] byteValuedDiscreteFeatures, short[] shortValuedDiscreteFeatures, float[] continuousFeatures, int setUnitIndex)

Method Summary

Methods

Modifier and Type	Method and Description
byte	getByteFeature(int index) An efficient way to access byte-valued features in this feature vector.
byte[]	getByteValuedDiscreteFeatures()
float	getContinuousFeature(int index) An efficient way to access continuous features in this feature vector.
float[]	getContinuousFeatures()
Number	getFeature(int index) A uniform way to access features in this feature vector.
int	getFeatureAsInt(int index) A wrapper to getFeature(), to get the result as an int value, e.g., for subsequent array indexing.
String	getFeatureAsString(int index, FeatureDefinition feaDef) A wrapper to getFeature(), to get the result as an String value, e.g., for subsequent System.out output.
FeatureVector.FeatureType	getFeatureType(int featureIndex)
int	getLength() Get the total number of features in this feature vector.
int	getNumberOfByteFeatures() The number of byte features in this feature vector.
int	getNumberOfContinuousFeatures() The number of continuous features in this feature vector.
int	getNumberOfShortFeatures() The number of short features in this feature vector.
short	getShortFeature(int index) An efficient way to access short-valued features in this feature vector.
short[]	getShortValuedDiscreteFeatures()
int	getUnitIndex() Get the index of the unit to which the current feature vector applies.
boolean	isByteFeature(int index) Test whether the feature with the given index number is a byte feature.
boolean	isContinuousFeature(int index) Test whether the feature with the given index number is a continuous feature.
boolean	isEdgeVector(int edgeIndex) Is this an edge vector?
boolean	isShortFeature(int index) Test whether the feature with the given index number is a short feature.
String	toString() Return a string representation of this set of target features; feature values separated by spaces.
void	writeTo(DataOutput out) Write a binary representation of this feature vector to the given data output.

Methods inherited from class java.lang.Object

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

Field Detail

unitIndex

public final int unitIndex

byteValuedDiscreteFeatures

public final byte[] byteValuedDiscreteFeatures

shortValuedDiscreteFeatures

public final short[] shortValuedDiscreteFeatures

continuousFeatures

public final float[] continuousFeatures

Constructor Detail

FeatureVector

public FeatureVector(byte[] byteValuedDiscreteFeatures,
             short[] shortValuedDiscreteFeatures,
             float[] continuousFeatures,
             int setUnitIndex)

Method Detail

isEdgeVector

public boolean isEdgeVector(int edgeIndex)

Is this an edge vector?

Parameters:

edgeIndex - edge index

Returns:

isEdge

getFeatureType

public FeatureVector.FeatureType getFeatureType(int featureIndex)

getLength

public int getLength()

Get the total number of features in this feature vector.

Returns:

the number of features, irrespective of their type

getUnitIndex

public int getUnitIndex()

Get the index of the unit to which the current feature vector applies.

Returns:

The related unit index.

getNumberOfByteFeatures

public int getNumberOfByteFeatures()

The number of byte features in this feature vector.

Returns:

a number of byte features, possibly 0.

getNumberOfShortFeatures

public int getNumberOfShortFeatures()

The number of short features in this feature vector.

Returns:

a number of short features, possibly 0.

getNumberOfContinuousFeatures

public int getNumberOfContinuousFeatures()

The number of continuous features in this feature vector.

Returns:

a number of continuous features, possibly 0.

getFeature

public Number getFeature(int index)

A uniform way to access features in this feature vector.

Parameters:

index - a feature index between 0 and getLength()-1

Returns:

a Number object, which will be a Byte, a Short or a Float depending on the type of the feature with the given index number.

getFeatureAsInt

public int getFeatureAsInt(int index)

A wrapper to getFeature(), to get the result as an int value, e.g., for subsequent array indexing.

Parameters:

index - A feature index between 0 and getLength()-1.

Returns:

The feature value, as an int.

See Also:

getFeature(int)

getFeatureAsString

public String getFeatureAsString(int index,
                        FeatureDefinition feaDef)

A wrapper to getFeature(), to get the result as an String value, e.g., for subsequent System.out output.

Parameters:

index - A feature index between 0 and getLength()-1.

feaDef - A FeatureDefinition object allowing to decode the feature value.

Returns:

The feature value, as a String.

See Also:

getFeature(int)

getByteFeature

public final byte getByteFeature(int index)

An efficient way to access byte-valued features in this feature vector.

Parameters:

index - the index number of the byte-valued feature in this feature vector.

Returns:

the byte value of the feature with the given index.

Throws:

IndexOutOfBoundsException - if index<0 or index ≥ getNumberOfByteFeatures(). check getNumberOfByteFeatures() . check isByteFeature(int index) .

getShortFeature

public final short getShortFeature(int index)

An efficient way to access short-valued features in this feature vector.

Parameters:

index - the index number of the short-valued feature in this feature vector.

Returns:

the short value of the feature with the given index.

Throws:

IndexOutOfBoundsException - if index<getNumberOfByteFeatures() or index ≥ getNumberOfByteFeatures()+getNumberOfShortFeatures().

See Also:

getNumberOfByteFeatures(), getNumberOfShortFeatures(), isShortFeature(int index)

getContinuousFeature

public final float getContinuousFeature(int index)

An efficient way to access continuous features in this feature vector.

Parameters:

index - the index number of the continuous feature in this feature vector.

Returns:

the float value of the feature with the given index.

Throws:

IndexOutOfBoundsException - if index<getNumberOfByteFeatures()+getNumberOfShortFeatures() or index ≥ getLength().

See Also:

getNumberOfByteFeatures(), getNumberOfShortFeatures(), getNumberOfContinuousFeatures(), getLength(), isContinuousFeature(int index)

isByteFeature

public boolean isByteFeature(int index)

Test whether the feature with the given index number is a byte feature.

Parameters:

index - index

Returns:

This will return true exactly if getByteFeature(index) would return a value without throwing an exception, i.e. if index≥0 and index < getNumberOfByteFeatures().

isShortFeature

public boolean isShortFeature(int index)

Test whether the feature with the given index number is a short feature.

Parameters:

index - index

Returns:

This will return true exactly if getShortFeature(index) would return a value without throwing an exception, i.e. if index≥getNumberOfByteFeatures() and index < getNumberOfByteFeatures()+getNumberOfShortFeatures().

isContinuousFeature

public boolean isContinuousFeature(int index)

Test whether the feature with the given index number is a continuous feature.

Parameters:

index - index

Returns:

This will return true exactly if getContinuousFeature(index) would return a value without throwing an exception, i.e. if index≥getNumberOfByteFeatures()+getNumberOfShortFeatures() and index < getLength().

getByteValuedDiscreteFeatures

public byte[] getByteValuedDiscreteFeatures()

getShortValuedDiscreteFeatures

public short[] getShortValuedDiscreteFeatures()

getContinuousFeatures

public float[] getContinuousFeatures()

writeTo

public void writeTo(DataOutput out)
             throws IOException

Write a binary representation of this feature vector to the given data output.

Parameters:

out - the DataOutputStream or RandomAccessFile in which to write the binary representation of the feature vector.

Throws:

IOException - IOException

toString

public String toString()

Return a string representation of this set of target features; feature values separated by spaces.

Overrides:

toString in class Object

Returns:

out.toString