I'm conversing with the judges over it, but here are my initial thoughts. Note that they are not a final ruling and we'll get back to you on it.
TTS, vocaloids, and similar voice simulation and production software should be acceptable with a valid license to use both the software and voice.
Two reasons:
1)TTS is often used as an accessibility feature and is actually built into some visual novel engines.
2)Vocaloids and the like have very well established licensing and rules regarding their usage. One of our major concerns with the current landscape of generative AI is the dubious nature of copyright, both in the source of the training material and in the ownership of the end product, and thankfully many soundfont packs have this well established.