After watching a bunch more streams I think making songs is not fully automated for Neuro. She sings only one new song every stream (and often repeats it). It's probably still fairly automated, that's why sometimes she makes strange noises or briefly sings in the wrong parts.
Neuro-sama
tries again. This is the only song they've felt a need to "fix." Still sounds poor. Actually I think her original version sounds better. This one seems tweaked and artificial. Not sure what exactly triggers the
growling. I like listening to her. Maybe being a robot grants her perfect pitch (I haven't heard vocaloid music much to compare).
QUOTE(Ass Spanker @ hoarse and hissy voice)

They won't but you will destroy your throat physically if you try to do those kinds of voice without proper training. I did, though I only ended up with a few days not being able to talk. Doing high pitched squeaky voice is really damaging. You can notice a few seiyuus changing their signature loli roles into more mature roles over the years if they debuted during their middle/high school.
Yeah I noticed that too, it's pretty standard.
Medium-high pitch voice is potentially not as damaging for me; I feel it's more a question of sounding the way I want, versus the quality and stability that comes out. I can choose from like 3+ kinds of squeaky voice in the range most songs are at (even many guy songs seem keyed in the female range).
Ghost Rule is almost comfortable for me in a Hatsune Miku voice, but I dunno how to make mafumafu voice, at least not in that quality.
Ironically, the person in the world with the voice closest to mine (maybe) turned out to be a hololiver that was born a month after I learned what a vtuber was. I think their voice is natural since I know mine is, but I don't know if they are a guy or a girl in real life. Uniquely, they refuse to sing anything (and never laugh, sneeze, or cough) so it's like they are hiding something. I'm not one of those suspicious dudes though so I have to pray she is a girl based on her looks.
The hololiver also likes to make deep guy voice and little sister voice, so she's perfect for me to compare with what I can generate. I first noticed our similarity from her "little sister" voice, which was almost the same as my "stuffed animal" voice, which I used to use often (only when no human can hear me). Just like me, she was refusing to sing even
happy birthday in a stream. The couple clips she sung, she uses the little girl voice (which is also what I'd often default to).
Later I noticed that her serious speaking voice is even more similar to my real voice, to the point that I mistake it for myself in certain clips and get major cringe.
I'm old though so I probably shouldn't push my voice at this age. This morning I woke up feeling better than usual, and I was listening to that Take on Me octave up by the professional girl. For some reason I felt that I could get close, and indeed I did (horrible screeching quality). I almost got up to E6 twice (probably D6 or so).
But then blood came out of my nose and now my throat is a bit raspy. (Wait, where did that blood originate?)
I could probably hit that with training, maybe in more than one way. Pointless though because I'll never be anywhere close to a real girl in quality. I think when I was younger I knew how to go an entire octave higher, but I forgot the feeling.
The guy from above actually explained it, but his explanation was the same as what I innately felt. "Just self-experiment, find and do it and suddenly go an octave higher" haha.
This also reminds me that my voice actually never changed in school. I read this year that girls' voices do deepen but only a few tones. I probably went the same. Some people in high school had remarked that my voice never changed.
There are also some other physical changes that didn't occur for me until I was 30, or never at all.Wait
this guy is awesome. Okay I will probably improve a lot from this guy. He has 100 videos of singing. However I must rest for 1 week to a month, after I bled from trying to scream a falsetto E6.