Acoustic Parameters Of Distinctive Features For Nasal Finals In Mandarin

Nasal finals was composed by vowels and nasal consonant . Because of the impact of Chinese dialect, nasal finals are more difficult to learn for many people in the study of pronunciation. And are prone to make errors and defects. So, Nasal finals are the focus of pronunciation evaluation. This work try to extract the distinguish parameters of nasal finals, to pronunciation evaluation as the background. the main jobs are as follows:First, this article analyzes the pronunciation, acoustic parameters, distinguish parameters and errors or defects in pronunciation evaluation of nasal finals. But it also has a certain gap between distinguish parameters in phonetics and practical application. So generate an idea of analysising distinguish parameters and common pronunciation errors by signal processing, and trying to find the relation of parameters, errors and signals.Second, we identify the core of speech samples in stand of pronunciation errors through analysising speech spectrum. We take advantage of Mel- Frequency Cepstrum Coefficient (MFCC) It has the score above 95% when we distinguish the confused nasal finals caused by pronunciation at the beginning. And the score above 90% when we distinguish the confused nasal finals caused by the main vowels. When we distinguish the confused nasal finals caused by nasal consonant , we employ the Freqrecon parameter. It has the score above 95%.Last, based on the analysising of distinguish parameters about nasal finals, we organized a framework of pronunciation classification about nasal finals. Basically completed the distinction of confused nasal finals. The framework is in accordance with the classification in Phonetics about nasal finals. Perhaps, it is useful to pronunciation evaluation about nasal finals from the perspective of acoustic parameters.