We would need a voice, and instructions about what it should say or do.

If your voice sounds different from some other voice, we ought at least have some generic modifiable voices or categories of voice, so we can categorise your voice as being either a voice of a certain category or a voice such as one might get using a certain set of filters/modifiers as input to a voice simulator.

Basically if voice synthesisers cannot yet produce a voice such as yours they need improving, and we want the open source version of such a voice-synthesiser.

We do not want a recording of you singing the song with your closed source human voice, we want the instructions as to how a voice is to be used to sing the song, maybe with some hints as to what kind of voice might perform it best for what kind of audience. (Maybe some audiences would prefer to hear it in a gruff voice, others in a soproano voice, others in a male voice, others in a femail voice, others in a unisex or roboti voice and so on and so on.)

We want source code for how to go about singing such a song.

-MarkM-