Generating phonemes with recognizers
Last reviewed: 12/2/2009
HOW Article ID: H120903
The information in this article applies to:
- GrammarKit 3
- LexiconKit 3
Recognizers may not always match because of varying pronunciation. Providing phoneme pronunciations can ensure accurate recognition results.
LexiconKit 3 provides a simple way to generate recognizer and synthesizer voice language-specific phonemes.
Phonemes are speech API (e.g., vendor) and recognizer/synthesizer voice (e.g., language) specific. They represent the basic element of speech sound used to aggregate speech.
The LexiconKit GeneratePhonemes method provides a simple way to generate the phonemes that can be edited to achieve the desired recognition or synthesis playback when using voices to generate them.
The following example illustrates generating phonemes with the VoCon 3200 V3 recognizer:
// Generate phonemes String stringVal = NChantLM1.GeneratePhonemes("apple", "Nuance VoCon3", ChantPartOfSpeech.CPOSNoun);
Refer to programming language specific syntax in the help file Class Library Reference.
When generation phonemes with VoCon recognizer, the language is determined by the current acoustic model.
When generating phonemes with Microsoft SAPI 5 recognizer in previous releases, the system default voice was used which may or may not have been the same language as the recognizer. In LexiconKit 3, the requested recognizer is used to generate the phonemes.