A Singing Voice Synthesizer Controlled by Arm Motions Using Compressed Phoneme Determination Algorithm
Masashi Ito, Tomohiro Ashina, Yujiro Saegusa · 2014
A singing voice synthesizer has been developed for an entertainment tool. Using the synthesizer, operator can control pitch, loudness, and depth of vibrato for the vowel-like output sound by his arm motions. The phonetic quality of the sound is controlled by a wrist angle of the operator. Since unnecessary formant transition often occurs in this phoneme determination, the improved algorithm is proposed on the basis of a spring-damper system in the present study. The algorithm consists of wrist angle prediction and refractory period utilization, which are derived from the analysis of the wrist angle movements. The simulation of phoneme determination supported both mechanisms to be effective.