Mandarin Speech Perception And Unit Pronunciation Quality Assessment

With the rapid development of technology,people’s needs and understanding of speech recognition and speech synthesis technology are also constantly improving,and speech technology is also affecting people’s lives.In all communication,correct listening and correct output have a very big role.The study of phonetics is also constantly expanding,so it is very meaningful to study the second language acquisition of the speaker.Firstly,using the methods of experimental phonetics and statistical probability,under the theory of perceptual assimilation model,20 students from different levels of the group were used to conduct perceptual experiments,and the learners’ perception accuracy,perceived response time and perception were examined.Distinguish between similar sounds.Secondly,based on the vowel features produced by 10 standard discourse speakers,the vowel data produced by 50 different levels of Putonghua learners are sorted and feature extracted,and the vowel acoustic space of the standard ordinary Hu discourse speakers is used.Compare and explore the characteristics of learners’ unit sound acquisition at different levels.Finally,the statistical analysis of acoustic features is introduced into the neural network.The deep neural network(DNN)is used to extract the features of the speech data to form the standard dissident acoustic model.The parameter matching method is used to calculate the pronunciation accuracy of the learners’ vowels and the total Accuracy of pronunciation.According to the results obtained by the learner’s speech perception experiment and the results of the output experiment and the acoustic accuracy of the learner’s pronunciation,the relationship between the learner’s vowel perception and the vowel output is obtained.Through the sensory experiment and output experiment of the two groups of learners,it can be known that the higher the language proficiency of the learner,the higher the perceived accuracy and the output accuracy.According to the vowel perception accuracy rate and the vowel output accuracy rate,it is concluded that the learner vowel perception affects the learner’s vowel output ability to a certain extent.This research provides teaching opinions for Chinese learners and broadens the application field of speech learning models.