Hi,
Since VSL is having fun with filters, I wonder if they can try something with a vocal formant filter.
One of the main missing features in choirs is multiplicity of phonemes. Sampling many of them is expensive and hard to manage. And you will never have enough phonemes to satisfy everybody. Take Fluffy's Dominus: there is an incredible amount of sounds, but English speakers are missing several of the most common sounds.
What about trying a different approach? Vocal formant filters can be so extreme to sound like a vocoder. But subtler things can be done, to make a vowel or a consonant sound a bit like the nearby one.
This would never result in a synthetic language, but for a singing choir it could approximate well enough the perception of words.
It could be just a slider: select the "Aaaah", and move the formant filter. It will smoothly move between a -> ɑ -> æ -> ɐ, maybe even reaching ə -> ɵ -> œ, and so on.
The same with consonants. A formant filter will let you transform a "p" in a "b", and a "t" in a "d", or a "k" in a "q" or a "g".
Paolo