Abstract:

This paper presents research on controlling synthetic voice via a MIDI accordion. As motivation for this
research served the goal of reviving "lost instruments", that is of using existing traditional instruments “lost” to new music genres as interfaces to control new types of sound generation in novel and future genres of music.
We worked with a MIDI-accordion, because it provides expressive control similar to the human breath through the mechanism of the bellows. The control interface of the two hands is also suited for characteristics of the singing voice: The right hand controls the pitch of the voice as traditionally used, while the left hand chooses combinations of timbres,
phonemes, longer phonetic units, and different kinds of vocal expression. We use a simple model of phonemes, diphones and timbre presets. We applied this to a combination of sample based synthesis and formant
based synthesis. Our intention is to extend the work to physical models. A further goal of this research is to bridge the gap between ongoing research in the domain of synthesis of the singing voice with that of experimentation in novel forms of artistic performance, by examining different ways of interpreting synthesized voices via the MIDI accordion.