Analysis of formant and pitch information for Spanish phonemes

This work describes the formant and pitch information for the phonemes of the Spanish spoken in Mexico City. Four isolated words and two utterances spoken by four males and two females were used as a data base. Contours for the first three formants and the pitch were computed in each case using a linear prediction technique. Comparison of the mean values of the formant frequencies to previously reported values showed interesting differences in several instances. In addition, transition data were used to define rules for parameter control of a speech synthesizer.