A primary study on voice color control

The purpose of this experimental study is to examine several factors of voice color control for a text-to-speech system and to get the control rules of the voice color which can be adapted to the TTS system. In order to find such rules from natural speech, we have studied glottal source parameters and frequency characteristics for several voice colors. In this work voice colors were catalogued as: deep, sonorous, thick, soft, harsh, high tone, husky, shrill, and weak voice. For the voice source model, the LF-model was used and for the frequency characteristics the formant frequencies, bandwidths, and amplitudes were used. These acoustic parameters were tested by multiple regression analysis to achieve the general relation between these parameters and voice color