edu.cmu.sphinx.linguist.language.ngram.trie

Class BinaryLoader

java.lang.Object

edu.cmu.sphinx.linguist.language.ngram.trie.BinaryLoader

public class BinaryLoader
extends java.lang.Object

Class that provides utils to load NgramTrieModel from binary file created with sphinx_lm_convert. Routines should be called in certain order according to format:

• verifyHeader
• readCounts
• readQuant
• readUnigrams
• readTrieByteArr
• readWords

Constructor Summary

Constructors

Constructor and Description
BinaryLoader(java.io.File location)
BinaryLoader(java.net.URL location)

Method Summary

Modifier and Type	Method and Description
void	close() Should be called when model reading finished
int[]	readCounts() Reads language model order and ngram counts
NgramTrieQuant	readQuant(int order) Reads weights quantation object from stream
void	readTrieByteArr(byte[] arr) Reads trie in form of byte array into provided array.
NgramTrieModel.TrieUnigram[]	readUnigrams(int count) Reads array of language model unigrams
java.lang.String[]	readWords(int unigramNum) Reads vocabulary of language model.
void	verifyHeader() Reads header from stream and checks if it matches trie header

Methods inherited from class java.lang.Object

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

Constructor Detail

BinaryLoader

public BinaryLoader(java.io.File location)
             throws java.io.IOException

Throws:

java.io.IOException

BinaryLoader

public BinaryLoader(java.net.URL location)
             throws java.io.IOException

Throws:

java.io.IOException

Method Detail

verifyHeader

public void verifyHeader()
                  throws java.io.IOException

Reads header from stream and checks if it matches trie header

Throws:

java.io.IOException - if reading from stream failed

readCounts

public int[] readCounts()
                 throws java.io.IOException

Reads language model order and ngram counts

Returns:

array of counts where ordinal number is ngram order

Throws:

java.io.IOException - if reading from stream failed

readQuant

public NgramTrieQuant readQuant(int order)
                         throws java.io.IOException

Reads weights quantation object from stream

Parameters:

order - - max order of ngrams for this model

Returns:

quantation object, see NgramTrieQuant

Throws:

java.io.IOException - if reading from stream failed

readUnigrams

public NgramTrieModel.TrieUnigram[] readUnigrams(int count)
                                          throws java.io.IOException

Reads array of language model unigrams

Parameters:

count - - amount of unigrams according to counts previously read

Returns:

array of language model unigrams, see NgramTrieModel.TrieUnigram

Throws:

java.io.IOException - if reading from stream failed

readTrieByteArr

public void readTrieByteArr(byte[] arr)
                     throws java.io.IOException

Reads trie in form of byte array into provided array. Size of byte array is computed from previously read language model specifications, see NgramTrie

Parameters:

arr - - byte array to read trie to

Throws:

java.io.IOException - if reading from stream failed

readWords

public java.lang.String[] readWords(int unigramNum)
                             throws java.io.IOException

Reads vocabulary of language model. Ordinal number of word stays for wordId.

Parameters:

unigramNum - - amount of unigrams

Returns:

array of strings - vocabulary of language model

Throws:

java.io.IOException - of reading from stream failed

close

public void close()
           throws java.io.IOException

Should be called when model reading finished

Throws:

java.io.IOException - if stream was corrupted