Publications of Jordi Bonada (co-researcher of the Vocaloid technology):
http://www.iua.upf.es/mtg/publicacio...ink=&urldocid=
In this Japanese blog entry:
http://akira-izumi.cocolog-nifty.com...loid_ab7e.html
it is said that among the publications by Mr. Bonada, following two thesises explain basis of Vocaloid technology well.
Spectral Approach to the Modeling of the Singing Voice
http://www.iua.upf.es/mtg/publicatio...001-bonada.pdf
Singing Voice Synthesis Combining Excitation plus Resonance and Sinusoidal plus Residual Models
http://www.iua.upf.es/mtg/publicatio...2001-celma.pdf
Too technical and mathematical for me.
The diagram from latter thesis shows the mechanism of Vocaloid: