Thank you Tom!
You are right, actually these voices are made to work good only in 1 or 2 octaves at maximum.
The one I posted simulates an opera singer voice (SOP), that is made for higher notes until its starts sound fake.
Lower notes don't even work, but the first and last sample is used in a wide range of the KB in order not to have NO SOUND keys.
The only way we could have more realistic voices would be having different samples for each note, that would result in a 70MB voice or more (not interesting), since one sample has about 1MB. This one has 14 samples only, concentrated to those 2 octaves it work better. The octaves lower and higher it is just changed the pitch of the first and last sample that makes it sould fake. It happens to most internal voices as you know.
That's why VSTs can be better than internal voices from keyboards, because the size is not a problem. We can find for example a very authentical piano voice with more than 2GB, where every note in all 7 octaves has a unique sample and also each note is multilayer that sounds different according the strenght you play.
I though yamaha version more unreal (sinthetic to my taste), but can be used depending on the song.
I also think they can be used only in specific situations, similar to ther TY similar voices.
Anyway, I'm happy with the result I got converting this voice from Symphony Of Voices VST limited to 13MB. There are some other good voices to be converted. I need to check one by one to see which are usable, but my time is short lately.
Did you check the other voice I posted? It sound very nice mainly for L part (neet to change octave to -1 or -2 to work correctly in L part)
Best regards!
Eduardo