/**
* If the given string ends with a consonant, insert a syllable boundary before that consonant.
* Otherwise, append a syllable boundary.
*
* @param s input syllable
* @return syllable with boundaries reset
*/
private String rebuildTrans(String s) {
AllophoneSet set = jphon.getAllophoneSet();
if (set != null) {
Allophone[] allophones = set.splitIntoAllophones(s);
if (allophones != null && allophones.length > 0) {
Allophone last = allophones[allophones.length - 1];
if (last.isConsonant()) { // insert a syllable boundary before final consonant
String lastPh = last.name();
return s.substring(0, s.length() - lastPh.length()) + "-" + lastPh;
}
}
}
return s + "-";
}
/**
* Converts an array of phone symbol strings into a single phone string. If stress is marked on
* input phone symbols ("1" appended), a crude syllabification is done on the phone string.
*/
public String phoneArray2phoneString(AllophoneSet allophoneSet, String[] voicePhones) {
StringBuilder phoneBuf = new StringBuilder();
for (int i = 0; i < voicePhones.length; i++) {
phoneBuf.append(voicePhones[i]);
}
Syllabifier syllabifier = allophoneSet.getSyllabifier();
return syllabifier.syllabify(phoneBuf.toString());
}
/**
* Convenience method to access the allophone set referenced in the MARY property with the given
* name.
*
* @param propertyName name of the property referring to the allophone set
* @throws MaryConfigurationException if the allophone set cannot be obtained
* @return the requested allophone set. This method will never return null; if it cannot get the
* allophone set, it throws an exception.
*/
public static AllophoneSet needAllophoneSet(String propertyName)
throws MaryConfigurationException {
String propertyValue = MaryProperties.getProperty(propertyName);
if (propertyValue == null) {
throw new MaryConfigurationException("No such property: " + propertyName);
}
if (AllophoneSet.hasAllophoneSet(propertyValue)) {
return AllophoneSet.getAllophoneSetById(propertyValue);
}
InputStream alloStream;
try {
alloStream = MaryProperties.needStream(propertyName);
} catch (FileNotFoundException e) {
throw new MaryConfigurationException(
"Cannot open allophone stream for property " + propertyName, e);
}
assert alloStream != null;
return AllophoneSet.getAllophoneSet(alloStream, propertyValue);
}
protected Map<String, List<String>> readLexiconStream(
String lexiconFilename, InputStream lexiconStream) throws IOException {
logger.debug(String.format("Reading lexicon from '%s'", lexiconFilename));
String line;
Map<String, List<String>> fLexicon = new HashMap<String, List<String>>();
BufferedReader lexiconFile = new BufferedReader(new InputStreamReader(lexiconStream, "UTF-8"));
while ((line = lexiconFile.readLine()) != null) {
// Ignore empty lines and comments:
if (line.trim().equals("") || line.startsWith("#")) continue;
String[] lineParts = line.split("\\s*\\|\\s*");
String graphStr = lineParts[0];
String phonStr = null;
try {
phonStr = lineParts[1];
} catch (ArrayIndexOutOfBoundsException e) {
logger.warn(
String.format(
"Lexicon '%s': missing transcription for '%s'", lexiconFilename, graphStr));
continue;
}
try {
allophoneSet.splitIntoAllophones(phonStr);
} catch (IllegalArgumentException e) {
logger.warn(
String.format(
"Lexicon '%s': invalid entry for '%s': %s",
lexiconFilename, graphStr, e.getMessage()));
continue;
}
String phonPosStr = phonStr;
if (lineParts.length > 2) {
String pos = lineParts[2];
if (!pos.trim().equals("")) phonPosStr += "|" + pos;
}
List<String> transcriptions = fLexicon.get(graphStr);
if (null == transcriptions) {
transcriptions = new ArrayList<String>();
fLexicon.put(graphStr, transcriptions);
}
transcriptions.add(phonPosStr);
}
lexiconFile.close();
return fLexicon;
}
