Package com.keenresearch.keenasr

Class KASRWordPronunciation

java.lang.Object

com.keenresearch.keenasr.KASRWordPronunciation

public class KASRWordPronunciation
extends Object

KASRWordPronunciation is a class that defines a mapping between a word and its phonetic pronunciation. A phonetic pronunciation is a space separated string of phonemes that define how the word is pronounced. The names of the phonemes are defined in ASR Bundle's lang/phones.txt file. For some languages, where this mapping is not deterministic, the ASR Bundle will contain a large lookup table in the lang/lexicon.txt file.

You can use KASRWordPronunciation to provide alternatives to existing pronunciations, or to define pronunciations for words that are not in the ASR Bundle lexicon. Some scenarios where this could be useful is modeling of: a) made-up words, b) common mispronunciations. The latter can be useful in use cases involving language learning and reading instruction. In those scenarios you can also provide an arbitrary tag value when constructing a WordPronunciation. The tag will be appended to the recognized word in the result, if such pronunciation was the most likely hypothesis. For example, given a word “PEAK” and tag “WRONG” the word in the result will be “PEAK#WRONG”.

Example: a pronunciation for PEAK that could potentially be tagged as WRONG is

 KASRWordPronunciation wordPronunciation = new KASRWordPronunciation("PEAK", "B IY1 K", "WRONG");
 

Constructor Summary

Constructors

Constructor	Description
KASRWordPronunciation(String word, String pronunciation)	Parameterized constructor.
KASRWordPronunciation(String word, String pronunciation, String tag)	Parameterized constructor.

Method Summary

Modifier and Type	Method	Description
final String	getPronunciation()	Get the phonetic transcription of a word, provided as a space-separated sequence of phones.
final String	getTag()	Get the optional tag value.
final String	getWord()	Get the string specifying the word.
boolean	isValid()	Verify if this object is valid, for a given KASRRecognizer.

Methods inherited from class java.lang.Object

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

Constructor Details

KASRWordPronunciation

public KASRWordPronunciation(String word,
 String pronunciation,
 String tag)

Parameterized constructor.

NOTE: both word and phones will internally be stored in uppercase.

Parameters:

word - string specifying the word. For example "CAT".

pronunciation - a space separated phone string.

tag - a tag suffix for the word.

KASRWordPronunciation

public KASRWordPronunciation(String word,
 String pronunciation)

Parameterized constructor.

NOTE: both word and phones will internally be stored in uppercase.

Parameters:

word - string specifying the word. For example "CAT".

pronunciation - a space separated phone string.

Method Details

isValid

public boolean isValid()

Verify if this object is valid, for a given KASRRecognizer.

A word pronunciation is valid if:

1. word does not contain < or > characters.
2. tag is not equal to "INC" (a reserved tag) and it does not contain a # character.
3. pronunciation is composed of valid phones (as defined in the ASR Bundle that was used to initialize the KASRRecognizer).

Returns:

true if this word pronunciation is valid; false otherwise.

getWord

public final String getWord()

Get the string specifying the word.

Example: "CAT".

Returns:

string specifying the word.

getPronunciation

public final String getPronunciation()

Get the phonetic transcription of a word, provided as a space-separated sequence of phones.

Returns:

space separated sequence of phones, provided during object creation. Phones will be in upper case and extra white spaces will be removed during object creation.

getTag

public final String getTag()

Get the optional tag value.

When provided during construction, this tag will be appended together with the # symbol to the word if the variation of the word with the provided pronunciation is recognized. For example, PEAK#WRONG.

Returns:

string specifying the tag.