Excitation pattern peak is more important determinant of vowel quality

The perceptions of five Chinese vowel /u, o, a, y, i/ and many perceptional phenomena can be explained well by the excitation pattern peaks. The study is reported in Science in China, Series F-Information Sciences, Volume 52，Issues 10 (Oct, 2009).

It is commonly accepted that the locations of prominent energy concentrations are related to vowel quality. But, because the spectrum of a sound undergoes many significant changes after entering the peripheral auditory system, no simple determinant relation is found between acoustic formant frequencies and vowel qualities.

This study investigates the relations between the phonetic qualities for five Chinese steady-state vowels and their excitation patterns, instead of acoustic formant. Both database measurement and auditory experiments demonstrated that each of the five vowels has only one representative peak on excitation pattern that determines the phonetic quality of the vowel, and other peaks, if any, do not have phonetic meaning.

This conclusion can explain many perception phenomena, such as "effective second formant" and "spectral integration". The study also shows that the representative peaks of the excitation pattern of vowel /u, a, i/ distribute along auditory frequency and are separate maximally, so they can be reliably distinguished from each other in the sense of hearing. It can explain why the three vowels occur most frequently in the languages of the world.

It is shown that in speech perception research the method based on inner representation is advantageous over those based on acoustic spectrum.

This research was funded by "211" Key Projects of Communication University of China.

Reference: YU ShuiYuan. Excitation patterns and phonetic identities of Chinese vowels. Science in China Series F-----Information Sciences 2009, 52 (10) 1936-1946 DOI: 10.1007/s11432-009-0164-z http://www.scichina.com:8083/sciFe/EN/abstract/abstract414649.shtml