En torno a las vocales del español: análisis y reconocmiento

Abstract

In this study we intended to carry out an analysis of male and femaIe vowel formants to obtain the data that would serve us to draw the boundaries between the different vowel categories, and could also be used for automatic recognition of vowels through association of any data outside the boundaries established by phonetic features. We have succeeded in proving that the first two vowel formants sufficed to characterize any of the five Spanish vowels, that F1 serves to distinguish between the features: high, middle and low, and also that F2 plays an important role in the identification of front, central and back vowels. We have aIso obtained two formuli that relate in a systematic way male vowel formants to female vowel formants. We have found the limits necessary (or the rules to work, and, finally we created a speech recognition programme and verified that 90% of the vowels of various words uttered by different speakers, were accurately recognised.