ACOUSTIC CUES OF FEMALE VOICE QUALITY.

In order to synthesize speech in accordance with a system of rules, the results of listening tests on synthesized sounds are applied to the various sound parameters for the female voice. The sound parameters studied included frequency locations and widths of peaks (or hills) of the spectrum envelope (not necessarily the same as formants), envelope slopes, and fundamental frequencies. In addition, an analysis was made of five vowels spoken by 34 males and 29 females, and a comparison was made of the listening-test results and the analytical data, which showed that the results of the listening tests were quite appropriate. A study was included of the rules for transforming the male voice into the female voice based on female speech synthesis from steady vowels, and the results were applied to a speech-synthesis system based on the vowel-consonant-vowel (VCV) phoneme chains of the male voice. These studies have shown that it is possible in this way to synthesize good-quality female speech.