Statistics of Analysis-by-Performance : Gesturally-Controlled Voice Synthesis as an Input for Modelling of Vibrato in Singing