edu.cmu.sphinx.result

Class WordResult

java.lang.Object

edu.cmu.sphinx.result.WordResult

public class WordResult
extends java.lang.Object

Represents a word in a recognition result. This is designed specifically for obtaining confidence scores. All scores are maintained in LogMath log base.

Constructor Summary

Constructors

Constructor and Description
WordResult(Node node) Construct a WordResult using a Node object and a confidence (posterior).
WordResult(Word w, TimeFrame timeFrame, double score, double posterior) Construct a word result with full information.

Method Summary

Modifier and Type	Method and Description
double	getConfidence() Returns a log confidence score for this WordResult.
Pronunciation	getPronunciation() Gets the pronunciation for this word.
double	getScore()
TimeFrame	getTimeFrame()
Word	getWord() Gets the word object associated with the given result.
boolean	isFiller() Does this word result represent a filler token?
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

WordResult

public WordResult(Word w,
                  TimeFrame timeFrame,
                  double score,
                  double posterior)

Construct a word result with full information.

Parameters:

w - the word object to store

timeFrame - time frame

score - acoustic score of the word

posterior - of the word

WordResult

public WordResult(Node node)

Construct a WordResult using a Node object and a confidence (posterior). This does not use the posterior stored in the Node object, just its word, start and end.

Parameters:

node - the node to extract information from

Method Detail

getScore

public double getScore()

Returns:

total score for this WirdResult in log domain

getConfidence

public double getConfidence()

Returns a log confidence score for this WordResult. Use the getLogMath().logToLinear() method to convert the log confidence score to linear. The linear value should be between 0.0 and 1.0 (inclusive) for this word.

Returns:

a log confidence score which linear value is in [0, 1]

getPronunciation

public Pronunciation getPronunciation()

Gets the pronunciation for this word.

Returns:

the pronunciation for the word

getWord

public Word getWord()

Gets the word object associated with the given result.

Returns:

the word object

getTimeFrame

public TimeFrame getTimeFrame()

Returns:

time frame for the word

isFiller

public boolean isFiller()

Does this word result represent a filler token?

Returns:

true if this is a filler

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object