A few thoughts after reading through this thread:
To be honest, I really like the idea of well-made, commercial UTAU voicebanks. Not everyone can afford Vocaloid, and I like the fact that, ultimately, development of voicebanks is fairly open-source. Yes, that means there's a drastic variation in quality from voicebank to voicebank, but I actually don't think that matters much. There's been a marked improvement over the last couple of years when it comes to voicebank quality; where Ritsu's KIRE voicebank was once a standout, multipitch voicebanks are now almost standard. The bar was raised, and will continue to be raised as long as the people who make voicebanks still care.
No, UTAU doesn't meet Vocaloid's quality level, but I feel like that doesn't matter so much, nor should it exclude UTAU voicebanks from being sold. Voicebanks are a tool -- the quality of the tool is important, but equally important is the skill of who's using it. There are plenty of terrible Vocaloid covers out there, made by people who just didn't care; at the same time, there are plenty of
amazing UTAU covers out there, made by people who
do care. To me, the existence of the not-so-awesome voicebanks detracts nothing from the awesome ones -- if anything, by comparison, the awesome banks shine even brighter for the quality of the work their creators poured into them.
As for all the discussion of free vs. commercial -- for me, the decision to buy a voicebank would come down to liking what that voicebank has to offer. I mean, I love Gahata Meiji and I love Kohaku Merry, but they are very different as far as voices go, and -- to me -- not really interchangeable. If I like the voicebank -- if I like the tone and quality of the voice, it's well-configured, the design is neat, and it's not too expensive -- I'd likely be interested in buying it (assuming I had disposable income for such a thing at the moment.)
I also kind of disagree that, with UTAU, there's no quality control, and thus, no guarantee of quality for commercial voicebanks. No, there isn't some regulatory committee deciding which banks can and can't be sold, nor who can and can't develop them (and that's a good thing, in my book.) However, while there isn't any real quality control on the development side, there
is on the end-user's side.
I can't imagine anyone buying a voicebank when they don't know what it sounds like -- that'd be like buying shoes without trying them on first. Even if you know those shoes are in your size, and you're familiar with the brand, there's no guarantee those particular shoes will feel right when you wear them. At the end of the day, few people are going to use a freeware voicebank that doesn't pass muster, and even fewer are going to pay for one that's subpar.
There's been some good ideas, I think, about what features developers might offer to make their voicebanks more appealing commercially (multipitch, solid configuration, flexibility as far as configuration format goes, physical releases.) Maybe there also needs to be some discussion about what those developers who are serious about going commercial can do to show the quality of their work, and so people contemplating purchasing a voicebank can see exactly what they'd be getting for their money.
Demo songs are good, but good mixing and tuning can obscure poor OTOing and vocal quality. Personally, I like it when voicebank developers use raw vocals as samples of vocal quality -- specifically, when they use raw vocals of the voicebank singing a scale. Maybe commercial developers could offer something like a limited-function demo, a voicebank using a handful of sounds per octave. Not complete enough to render off a song, but enough for prospective buyers to get an idea of what the voicebank offers in the way of vocal clarity, OTOing quality, etc., without the voicebank developer having to give an entire bank away for free.
So.... yeah. Going back to lurking now!