inpro.sphinx.decoder
Class FakeSearch

java.lang.Object
  inpro.sphinx.decoder.NoSearch
      inpro.sphinx.decoder.FakeSearch

All Implemented Interfaces:

edu.cmu.sphinx.decoder.search.SearchManager, edu.cmu.sphinx.util.props.Configurable

public class FakeSearch
extends NoSearch

recognizing is so time consuming why not have a recognizer that doesn't recognize at all, but fakes results from a given transcription, just consumes frames from the frontend and calls event listeners and monitors as needed

Author:

timo

Nested Class Summary

private class

FakeSearch.TokenList

Field Summary

private edu.cmu.sphinx.linguist.dictionary.Dictionary	dictionary
private int	frameNumber
private static org.apache.log4j.Logger	logger
static java.lang.String	PROP_DICTIONARY
static java.lang.String	PROP_TRANSCRIPT_FILE
static java.lang.String	PROP_UNIT_MANAGER
static java.lang.String	PROP_UNIT_TIER
static java.lang.String	PROP_WORD_TIER
private FakeSearch.TokenList	sortedTokenList
private java.lang.String	transcriptName
private edu.cmu.sphinx.linguist.acoustic.UnitManager	unitManager
private java.lang.String	unitTierName
private java.lang.String	wordTierName

Fields inherited from class inpro.sphinx.decoder.NoSearch

emptyResult, fe, PROP_FRONTEND

Constructor Summary

FakeSearch()

Method Summary

void	allocate()
void	loadTranscript(java.lang.String filename)
void	loadTranscript(TextGrid tg) create the internal sortedTokenList from the given Textgrid this works as follows: - iterate over the words in the word tier - find pronunciation from the dictionary - generate a PronunciationState and Token for this word and add it to the sortedTokenList - for each unit in the pronunciation - assert that the given unit labels match the units in the pronunciation - generate a UnitState and Token for this unit and add it to the sortedTokenList
void	newProperties(edu.cmu.sphinx.util.props.PropertySheet ps)
edu.cmu.sphinx.result.Result	recognize(int nFrames) consume the given number of frames from the frontend, find the Token for the result (completely ignoring the data...)
void	setTranscript(java.lang.String filename)

Methods inherited from class inpro.sphinx.decoder.NoSearch

deallocate, startRecognition, stopRecognition

Methods inherited from class java.lang.Object

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

Field Detail

logger

private static final org.apache.log4j.Logger logger

PROP_DICTIONARY

@S4Component(type=edu.cmu.sphinx.linguist.dictionary.Dictionary.class)
public static final java.lang.String PROP_DICTIONARY

See Also:

Constant Field Values

PROP_UNIT_MANAGER

@S4Component(type=edu.cmu.sphinx.linguist.acoustic.UnitManager.class)
public static final java.lang.String PROP_UNIT_MANAGER

See Also:

Constant Field Values

PROP_TRANSCRIPT_FILE

@S4String(defaultValue="")
public static final java.lang.String PROP_TRANSCRIPT_FILE

See Also:

Constant Field Values

PROP_WORD_TIER

@S4String(defaultValue="MAUS-words")
public static final java.lang.String PROP_WORD_TIER

See Also:

Constant Field Values

PROP_UNIT_TIER

@S4String(defaultValue="MAUS-phones")
public static final java.lang.String PROP_UNIT_TIER

See Also:

Constant Field Values

dictionary

private edu.cmu.sphinx.linguist.dictionary.Dictionary dictionary

wordTierName

private java.lang.String wordTierName

unitTierName

private java.lang.String unitTierName

transcriptName

private java.lang.String transcriptName

sortedTokenList

private FakeSearch.TokenList sortedTokenList

frameNumber

private int frameNumber

unitManager

private edu.cmu.sphinx.linguist.acoustic.UnitManager unitManager

Constructor Detail

FakeSearch

public FakeSearch()

Method Detail

allocate

public void allocate()

Specified by:

allocate in interface edu.cmu.sphinx.decoder.search.SearchManager

Overrides:

allocate in class NoSearch

recognize

public edu.cmu.sphinx.result.Result recognize(int nFrames)

consume the given number of frames from the frontend, find the Token for the result (completely ignoring the data...)

Specified by:

recognize in interface edu.cmu.sphinx.decoder.search.SearchManager

Overrides:

recognize in class NoSearch

Parameters:

nFrames - the number of frames to recognize

Returns:

a result that contains information from the transcription up to the current time

loadTranscript

public void loadTranscript(TextGrid tg)

create the internal sortedTokenList from the given Textgrid this works as follows: - iterate over the words in the word tier - find pronunciation from the dictionary - generate a PronunciationState and Token for this word and add it to the sortedTokenList - for each unit in the pronunciation - assert that the given unit labels match the units in the pronunciation - generate a UnitState and Token for this unit and add it to the sortedTokenList

Parameters:

tg - given TextGrid

loadTranscript

public void loadTranscript(java.lang.String filename)

setTranscript

public void setTranscript(java.lang.String filename)

newProperties

public void newProperties(edu.cmu.sphinx.util.props.PropertySheet ps)
                   throws edu.cmu.sphinx.util.props.PropertyException

Specified by:

newProperties in interface edu.cmu.sphinx.util.props.Configurable

Overrides:

newProperties in class NoSearch

Throws:

edu.cmu.sphinx.util.props.PropertyException