Vowel Experimental And Computational Study

Speech sound is the physical media of oral communication. It was studied early. And the important ancient production was the creation of letter. Nowadays the present study of speech sound has become a branch of linguistics which is similar to science. Not only instruments are used in studying speech sound but also the knowledge of linguistics, physiology, physics, psychology and sociology is needed. On the other hand, the productions of studying speech sound are applied in studying linguistics, teaching languages, hearing therapy, artificial intelligence. This paper focuses on vowel. The calculating method is studied, which is of perceptive pattern on psychological level based on the speech sound system theory. And the problem of individual differences of vowel formants among different persons is solved, which will help the investigation of languages, the sound change of languages and so on.In Chapter One, the study of vowels in the phonetic domain is summarized, including the study of physiological phonetics, the study of acoustic phonetics, the study of perceptive phonetics. And many concepts are discussed for new understanding.In Chapter Two, the speech sound system is theorized. There are three levels in it, which are physical level containing physical sound, psychological level containing psychological sound, sociological level containing sociological sound. The psychological level is the hinge, where the physical sound is transformed to psychological sound, thus the individual difference is removed.In Chapter Three, the experimental method and calculating method are discussed. A formants normalizing calculating method is put forward. The method is based on the vowel sound system and matrix theory. The concrete arithmetic comes from K-L Transform Theory. The detailed arithmetic method and examples are provided. And the merits are discussed.In Chapter Four, the vowels in Shanghai dialect are studied using the normalizing method to calculate formant frequency. The data from twenty-aged group and sixty-aged group are compared. There are 16 items changed from 172 items. The annual variability is 0.23%.