Research On The Method Of Chinese Vowels Pronunciation And The Causes Of Stuttering Based On DIVA Model

In recent years, the main aspect of the artificial speech synthesis system to pursuit is thedescription and simulation of the human brain function involved in speech understanding andgeneration region in neuropsychology and neuroanatomy level. Around this main idea, manyresearchers make a huge effort in the exploration and researching of speech acquisition andgeneration computing model, and gain fruitful results. One of the most representative andgroundbreaking one is DIVA (Directions Into Velocities of Articulators) model researched byprofessor Gunther (Frank H. Guenther) and his team of speech lab at Boston University. Bycomparing with other models, it can be said that testing and definition of the DIVA model isthoroughest among today’s speech acquisition and generation model which have biologicalsignificance, and it is a kind of model that applying pseudo-inverse control scheme.DIVA (Directions Into of Articulators) model is an adaptive neural network which is used tocontrol the movement of analog vocal tract for generating the words, syllables or phonemes. Thevoice background that DIVA model depends on is the29phonemes of English pronunciation. Thisarticle uses six vowels a, o, e, i, u, üas experimental carriers for exploring the possibility of theDIVA model to deal with the Chinese pronunciation mechanism, thereby lays a good foundation forprocessing all Chinese phoneme phonemic including consonants, ultimately achieves the purpose ofbuilding the speech recognition synthesis system of Chinese brain mechanisms reaction system onthe basis of the DIVA model and "reading" out the thought processes of the brain of Chinese.The article firstly describes the model, then theoretically discusses the basic approaches of themodel for dealing with the Chinese vowel and the similarities and differences for processingEnglish phonemes, finally simulation experiments for the pronunciation of these vowels are made.Experimental results show that after modifying the pronunciation dependent background of theDIVA model, it can correctly produce Chinese vowel sounds and pronounce well. Theseexperiments provide the experimental basis for the construction of Chinese Large VocabularyContinuous Speech Recognition (LVCSR) system future.Based on the DIVA model’s feedforward and feedback control mechanism, this article alsoinvestigates the causes of stuttering, and uses the DIVA model simulating stuttering pronouncedprocess.Researches done in the paper can make a significant contribution to improve the validity and reliability of the DIVA model, promote the development of neuropsychology and neuroanatomy,and accelerate research on medical devices.