Looking in the back of the manual on pages EN-351 and EN-352, Casio lists copyrights for five open source libraries for vocal synthesis and vocal singing. A considerable amount there lists work done at the Nagoya Institute of Technology, and the Tokyo Institute of Technology. There are web links there if you would like access to the code.
One of the copyrights is for Sinsy, ( wiki page on Sinsy ), which takes a MusicXML file and generates a WAV file. Supposedly there is an on-line version of Sinsy, to create singing audio. Sinsy only supports Japanese and English, though there is a Mandarin version as well.
ADDED : Someone with the skill and with a significant amount of work could probably port some combination of this code to run on a computer like a Raspberry Pi.