Unpopular Opinions

[Prism]

I think quality assurance is very important but even with vocaloid's long testing a lot slips through. I wish Gumi English had a proper update instead of the custom dictionary that Internet put out to fix her vowels. Or I wish tonio worked with the @ phoneme with out crashing the software. Or even Daina being able to say smile correctly. I think it hurts English vocaloids the most because of how complicated they are. I think synth v has done a great job so far with their updates they were quick to fix the maki English ai glitch and rikka's pronunciation. Native instruments does it too so it's not uncommon for Yamaha.

 

[___]

I'm surprised to hear you say the Meikas are LQ! In my experience I would've placed them to be around more MQ--the only thing I've noticed that I find genuinely bothersome about them is that Mikoto can sound a bit stopped up/stuffy nosed/singing like she has a sock stuck in her mouth at times. Hime's a bit breathy but it never detracted much for me! I've never really found them to be difficult to work with, and shockingly enjoyed working with them more than any Vocaloid I own aside from Gumi!! Mikoto (I can't vouch for Hime as I've yet to attempt this with her) is about medium difficulty to mix as well.
I'm curious about what you have to say about them!! Spill that tea

Also, an unpopular opinion to keep us on track:
I think SynthV as an editor is good, but its engine is needing improvements. I can't quite place my finger on what is about the engine that's a turn-off, but I feel like all the voices sound kind of... stiff, I guess? It surprises me considering the developer, Kanru Hua, also made Moresampler which is the UTAU resampler I swear by.

I also feel like Mikoto is very choppy and has very unpleasant rattling and inconsistent tone to her voice which makes her rather LQ imo and also both of their samples appear rather noisy and not as clear compared to many vocaloids

And yeah I do def agree with you on svs, I feel like there's a pressure in the fandom to like svs cuz everyone wants it to be the new vocaloid and it's quality gets overestimated when on average a svs vb resembles upper MQ noisy utau, which isnt just terrible cuz utau is a good engine but still.

 

[ruffe]

I will admit that sample-based SVS banks feel like they're only a few steps above being commercial UTAU - with the Doppeltler resampler now available it feels like UTAU like Sekka Yufu could stand amongst them. I think it'd benefit some of the people to realise that it's unlikely any commercial synth will ever beat Vocaloid in the sample-based department in terms of quality and likeness to the provider, and it's totally okay!

The only thing holding UTAU back from standing with the commercial synths is its lack of solid parameters I feel! though it's completely capable without them.

 

[Scarlet Illusion]

I am SO glad I'm not the only one who felt that way about svs. Also, I strongly agree with Luwo's statement about UTAU--to elaborate my thoughts, I think UTAU could benefit from more parameters (brightness would be really nice) as I find most of the flags to be kinda... useless? And, I think more phonetic control would be nice (like how I can edit Japanese vocaloid's phonemes to have them sing only the consonant, or how SynthV lets you adjust the timing of individual phonemes). Hopefully one day Ameya will give UTAU a proper update and overhaul to add in the few things that still seem to be missing.

 

[___]

pat pat Kazumi you aren't in this alone jhbgvf

And, I think more phonetic control would be nice (like how I can edit Japanese vocaloid's phonemes to have them sing only the consonant, or how SynthV lets you adjust the timing of individual phonemes

Oh that's all about the reclist! CVVC format allows for this even in UTAU and synths like Vocaloid and SVS are mainly CVVC-like hence theyre capable of it but since VCV is the standard japanese reclist for UTAU we're less used to it.

 

[peaches2217]

I will admit that sample-based SVS banks feel like they're only a few steps above being commercial UTAU - with the Doppeltler resampler now available it feels like UTAU like Sekka Yufu could stand amongst them. I think it'd benefit some of the people to realise that it's unlikely any commercial synth will ever beat Vocaloid in the sample-based department in terms of quality and likeness to the provider, and it's totally okay!

As an SVS stan, may I just say: hella. You can get smoother results more easily in SVS than in Vocaloid, but Vocaloid's still top of the game in terms of quality. I can see SVS reaching that standard and even surpassing it in a few years' time, but is it there now? No. That's no reason to discredit it, of course, because for how young a software it is it's amazing. But it's not Vocaloid-level HQ yet.

Oh that's all about the reclist! CVVC format allows for this even in UTAU and synths like Vocaloid and SVS are mainly CVVC-like hence theyre capable of it but since VCV is the standard japanese reclist for UTAU we're less used to it.

Yup yup! VCV is (iirc) a lot easier to record and to use. CVVC gives you much more control over things like phonemes and, to an extent, tone, but it always goes waaaaaaaaay over my head.

 

[___]

CVVC is easier to record but I think VCV is the easiest and most convenient way to use ( realistic/smooth ) japanese utaus hence it became standard:0c

 

[TheStarPalace]

I wish VCV was easy to record, I would make nothing but CVVC for the rest of time if I was good at using it

 

[___]

Sameeeeeeee i use VCVs mostly but only make CVVCs jnhgfdsvfcd

 

[cafenurse]

i see people complain sometimes that eng voicebanks "over pronounce" words which makes them sound unnatural, but honestly i really like that. i think it sounds nice and sometimes it gives me asmr.
but maybe thats just cause i over pronounce words, darn living in the midwest for two years and picking up the midwestern accent

 

[peaches2217]

i see people complain sometimes that eng voicebanks "over pronounce" words which makes them sound unnatural, but honestly i really like that. i think it sounds nice and sometimes it gives me asmr.
but maybe thats just cause i over pronounce words, darn living in the midwest for two years and picking up the midwestern accent

On a similar note, people reeeeaaaaally blow out of proportion just how difficult it is to adjust it so that they don't over-pronounce. I mean, it's not instantaneous, but there's a readily available list of English phonemes for Vocaloid on the wiki, and editing a word takes maybe ten seconds? And often times you don't even need a phoneme change, just an adjustment to the velocity. Given how ridiculously complex the English language can be, the fact that many English synths sound the way they do right out of the box is honestly quite impressive.

 

[Prism]

Honestly I rather have over pronunciated like the cybers than under. The open perimeter exists for a reason. I do think forced banks like maki English standard and Miku English can be harder to work with.

 

[Blue Of Mind]

I honestly don't think Luka is suited to hard rock and metal songs. Certainly no offense to producers who use her for those kinds of songs (and indeed, there are some Luka rock and metal songs I really love anyway), but I feel Luka gets drowned out by the instrumentals because her voice is so soft. If Luka had a proper Power VB, I think this issue would be less common, but instead we have EVEC for her V4 which is... not as good as Miku and the Kagamines' standard appends, to put it politely. (It's an interesting experiment, I admit, but normal VBs would have been better for Luka in the long run). If we're just looking at the Cryptonloids, Rin is the better rock/metal vocal, and among Vocaloids in general, the likes of Flower and Una Spicy can handle louder songs better than Luka.

 

[sketchesofpayne]

I honestly don't think Luka is suited to hard rock and metal songs.

I want to disagree... but it's kinda true.

 

[DefiantKitsune]

I'd say Luka suits rock and lighter-ish metal well enough BUT you have to know what you're doing when you mix it (which the better known Luka rock producers do, but a lot of others don't).

 

[Leon]

Fucking around with the phonemes on Engloids is imperative, because of the overpronunciation. It's just that they sound unnatural and even difficult to understand when you let them pronounce every single sound as if they were saying the word as properly as possible, because... well, people don't really talk or sing like that. The vast, vast majority, anyway. And editing phonemes isn't difficult, as Peaches said, although each bank tends to have its quirks to work around. It's fun!

RE: the Meikas, a friend of mine had them and gave me his opinion. They are just,, not great. Mikoto's transitions are rough and they're just choppy in general. Hime's samples sound like they had a breathy filter added over them, and some of their vowels will do that audible looping that old UTAUs do when you drag the note out. The Meikas are fine voicebanks and I enjoy them both, but I wouldn't call them high quality, especially not as V5s :S

VB quality stuff

Also chiming in on a topic already dropped, even back with old physical game releases there were still patches semi-often. For example, I've gotten into some older games here recently (Doom and Arena on DOSBox mostly) and Arena had a number of bugs and such that had to have patches put out for it. It's got a whole list of versions and bugs with different versions on the wiki LOL.

All this to say, updating voicebanks should have always been possible imo, and I don't personally think there's been a drop in synth quality since it became more accessible. If we're really holding Vocaloid as the paragon of quality here, the V2 Engloids, Arsloid, Unity-chan, and a good handful of others would like to have a word with you. And no one @ me for this, yall know I love the V2 Engloids, and I think Unity-chan and Arsloid are both adorable banks even if they're not what I'd call HQ,,

 

[ruffe]

Arsloid's main voicebank is on par with any other HQ bank. His extra voicebanks, though? Yeah I can't defend those--though they aren't even supposed to be used outside of XSY so I don't even count them as "proper" voicebanks, especially in V5 where appends are less important than ever.

 

[___]

Just how possible patches ( in this form ) would've been is honestly questionable seeing as any form of updates that happened during V1-2 era were pretty clunky, awkward and cauzed inconveniences.

As for unpopular opinion! I honestly don't mind more cutesy voicebanks! And I feel like this fandom needs to chill in regards to them too, the way the western side's treated Rosa just because of the prospect of her being cutesy is honestly sad.

 

[Scarlet Illusion]

I like cutesy voices as well--I think every voice pretty much has its place and appeals to a different audience. I understand being disappointed that "omg they made ANOTHER moe voicebank", but I don't understand people who want to whine and complain and otherwise act poorly about it. Okay... I might be disappointed as well when something's not quite what I wanted, but I'm not going to be rude about it. :/ (I don't particularly have any strong feelings for or against cutesy sounding voices, tbh.)
(Also, there are dozens of UTAU out there with a variety of voice types... why not just... find an UTAU with your desired voice type...? I know it would probably require learning a new synth, but if you want it THAT badly, I mean... c'mon....)

 

[peaches2217]

I like cutesy voices as well--I think every voice pretty much has its place and appeals to a different audience. I understand being disappointed that "omg they made ANOTHER moe voicebank", but I don't understand people who want to whine and complain and otherwise act poorly about it. Okay... I might be disappointed as well, but I'm not going to be rude about it. :/ (I don't particularly have any strong feelings for or against cutesy sounding voices, tbh.)
(Also, there are dozens of UTAU out there with a variety of voice types... why not just... find an UTAU with your desired voice type...? I know it would probably require learning a new synth, but if you want it THAT badly, I mean... c'mon....)

Something I've noticed: if a feminine vocal is anything other than mature and sultry and husky, it's often labeled "moe" without further discussion. Hippi's a good example. A lot of people were expecting, well, Lily 2.0, and she instead has a very youthful-sounding voice. I think it's much more realistically youthful than outright moe, but the number of people decrying her for being "uwu kawaii moe desu" and crap like that's just annoying. I wish people wouldn't see feminine vocals as so black and white.