A Model of Coarticulation

概要

論文の詳細を見る
In the study of speech, it is one of the principal problems to investigate the mechanism of context in speech and to express it by rule. We have constructed a simple rule for the transition of vocal tract area functions. The rule assumes target configurations S_1,S_2,... for each phoneme, and express the area function S on the way of transition between S_1 and S_2 as follows : S=S_1+(S_2-S_1)・T(n) (n=normalized time, T(n)=normalized locus of second formant frequency F_2 between S_1 and S_2). For the transition among S_1,S_2 and S_3,the locus of F_2 is assumed as the product of the locus of F_2 between S_1 and S_2 (T_1(n)) and the one between S_2 and S_3 (T_2(n)), and on the articulatory level it is expressed as follows : S=S_1+[S_2+(S_3-S_2)・T_2(n-n_a)-S_1]・T_1(n) (n_a=instant of beginning to carry out the command to move on to S_3). It has been shown that the results calculated by our rule coincide good enough with the observed effects on V_1 V_2 V_1,V_1 C V_2 and CVC context (V=vowel, C=stop or nasal) and that the results are available for phoneme discrimination.
山梨大学の論文