Falsetto is used by the AI engine based on the pitch of the note, and will be different for each voice database. There’s currently no way to force a falsetto tone with AI voices (though some vocal modes will definitely help).
As for the transition at 0:14, it sounds like a very brief increase in pitch, and probably a decrease in voicing. These sort of “voice crack” effects are possible with most voice databases. You can also hear the voicing parameter used elsewhere in the song like the whispered line at 0:42.
It’s also important to note that Creuzer is well known for their extremely detailed tuning, they likely spent many hours to get this to sound the way it does. Even if you know the techniques it’s difficult to implement them as well as Creuzer does without a lot of practice.
There’s no real way to know for sure, but honestly it doesn’t really matter.
From a detailed tuning perspective it’s almost always better to draw the parameter curves by hand, especially if you’re as experienced as Creuzer is. I don’t think they’d have any reason to rely on scripts.
Scripts can work as helpful shortcuts, but they’re rarely the best way to do things. If an experienced tuner wants a specific transition or vibrato, there’s no reason they wouldn’t just draw it manually.
It’s much more difficult and time-consuming to record yourself singing the exact way you want (which would likely take multiple attempts) and then apply that curve to SynthV.
The reason many inexperienced users tend to rely on things like instant mode, real voice, vocalshifter, etc. is because it takes a lot of practice to know what makes a vocal “realistic” in the first place. It’s often easier and faster to get an acceptable result by using those tools, but it doesn’t really compare to a skilled tuner making detailed moment-to-moment manual changes.
In the end the main difference is whether people know what they want a phrase to sound like, or if they’re relying on the unconscious things they do as part of a vocal performance without actually being aware of which individual factors make up that performance.
And to be clear, this isn’t a bad thing! These tools are significant time-savers and not everyone wants to spend hours fine-tuning a track. They’re just two different approaches to using the software.
This is the answer I’ve been looking for when it comes to scripting. I’m on a productivity ride now. I’d rather spend more time making sure my songs are well-written than fiddling with SynthV, because I think a passable singing performance on well-written songs will go further than an amazing performance on a bad song. So far, auto-tuning with about an hour of adding note transitions and making sure they’re all in tune has been able to make me happy with the results.
But since I used to be a (bad) singer, I’m super curious about the level of Cruezer detail, especially on Fly Me to the Moon.