Is it unlikely to make a computer sing ? Well does robots dream of electrical sheep, and can we not all sing in our own way. The plug in Cantor may not answer all our questions, but it might challenge what we know, or take for granted.
When I first heard of it, I was in a some sort of no mans land, could it sing the parts, that my own voice would not, or was it just a new computer speech program, like in the old Amigas OS. Well one thing it does not, is sounding like a computer talkie. Either it does sing in a way, or it sounds like a synthesizer of a whole new breed. It all depends on what you can fill in to it. Take my word for it there is not much of a preset library, there is one phoneme set, and that is it, well until further notice that is. But do not believe that to be a bad thing, because that leaves room for your creativity, and what a sphere it leaves you.
Here is what you get, a kind of an additive oscillator, Cube style, with a certain amount of partials, (of which turning the number up or down does not affect the cpu consumption, or so it seems, as far as I could find out from tweaking it, that is.) A default Phoneme set, and 16 user sets, Which maybe, maybe not, is going to be expanded later, ( I do not want to promise something that I do not know for sure. ) A voice editor, with gender setting, brightness, and various other factors for voice creation, twelve in all, but there are many settings more. Four Simultaneous effects, of which the reverb, ranging from Taj Mahal to Shoe Box, are outstanding, the reverb must be said though, like in Cube are a global effect, in an multitimbral 8 part, monophonic, synthetic singer. I can not mention all of the settings possible, but it is of course most important what results you can achieve with it, what the sum of this settings are, coming next.
It sounds like there is nothing wrong with the Letters a, e, o…. and so on but when we come to the b, c, d… breed they do not cut right through the air, in other words, they are not sounding loud and clear enough to be easily heard. But this might not be a problem if you make your own phoneme sets, I have actually been able to make it sing naaaa …..naaa…. naaa…Without being able to clearly distinct it from a girl singing these ahem …Vocalize these words, and with a very clear nnnh sound, but my guess would be that this used with only one word for each part would be most giving, and would be best for atmospheric non lyric music like New Age or Ambient.
But now I must mention that I have made it sound like an harmonium, and violin alike sounds, and I think there is much more in it than just that, flutes, maybe sitars, maybe not lifelike but close enough (or things maybe newer heard before) And do remember making new sets of phonemes these can really bring a lot of life to this Synthetic singer.