We need to start talking about the future of UTAU's engine development before it's too late

[partial]

Who is that person? I've never heard of them and I'm curious.

 

[chunter]

 

[KNΞMΛTCS]

Looks like it's a continuation/fork of QTAU. Nice.

 

[Cheese]

There's been some talk in the Japanese fandom over a new vocal synth named SiVo. The developer made a blog post mentioning how VB6 support is ending in 2024 (the programming environment UTAU is written in) and how he's making this essentially due to that.

Many users have been tweeting about this and the possible end of UTAU once VB6 support expiresーand there's even a word for it nowーUTAU終了説 (UTAU Termination Theory).

Spoiler

Personally, I think the more vocal synths we have, the better. However, many people in the Japanese fandom aren't really taking to it being branded as an UTAU replacement and the whole panic that's been caused by this. I agree with that sentiment 100%.

semi off topicーbut I had actually emailed Ameya in regards to some things for the UTAU zine I'm organizing, and I was surprised to get a reply from him last night. Ameya still hasn't stated UTAU's development status, so if someone asked him he might reply...?

edit: I asked him about it in a followup email regarding zine stuff, so let's hope he'll reply.

 

[Info-Chan]

There's been some talk in the Japanese fandom over a new vocal synth named SiVo. The developer made a blog post mentioning how VB6 support is ending in 2024 (the programming environment UTAU is written in) and how he's making this essentially due to that.

Many users have been tweeting about this and the possible end of UTAU once VB6 support expiresーand there's even a word for it nowーUTAU終了説 (UTAU Termination Theory).

Spoiler

Personally, I think the more vocal synths we have, the better. However, many people in the Japanese fandom aren't really taking to it being branded as an UTAU replacement and the whole panic that's been caused by this. I agree with that sentiment 100%.

semi off topicーbut I had actually emailed Ameya in regards to some things for the UTAU zine I'm organizing, and I was surprised to get a reply from him last night. Ameya still hasn't stated UTAU's development status, so if someone asked him he might reply...?

If you can provide his email i'll email him about it.

 

[Cheese]

If you can provide his email i'll email him about it.

I actually asked him about it in a followup email regarding other things—and forgot to edit my post, sorry about that.

 

[na4a4a]

I'm glad someone has taken notice to the fact that VB6 is dead as holy hell and decided to take action.

While Utau with eventually die due to obsolescence I hope that what replaces it is more or less a drop in replacement and fully compatible.

VB6 support technically already ended but it's left in Windows as a legacy feature. If anything users will be able to manually install it one way or another and keep beating the dead horse keep the dead horse beating.

 

[Nohkara]

About that "SiVo" thing that @Cheese mentioned:

To be really honest, I'm very skeptical of whole project at all. Ok, cool, the person behind it has realised that how VB6 support is ending in 2024 and because that they tweeted that "You cannot use UTAU anymore in 2024" - I remember, many MANY people in Twitter side get panicked!

Then they tweet that

"Because we cannot use UTAU in 2024 anymore, I introduce new program called SiVo that will replace UTAU. It's based on HMM coding."
"SiVo will support Japanese, English, Chinese, Korean *list continues* (...) for Japanese a talk voice too."
"I'm looking for female and male voicer for each language"

"To create a "voice bank" data for SiVo you will need to SING about 50 to 100 different songs" (yes, they told that VB will be made of actual REAL singing samples!) and them many people at Twitter have been like "but what if voicer is really bad at singing and cannot sing at all???" and "will UTAU voicebanks been supported in SiVo at all???" and after many this messages answer has been for that something like "will make support UTAU VB some way".

And just recently someone suggested that "what if you arrange a design contest for SiVo's coming "default voice"?" and of course they agreed and started it right away without giving contest rules or deadlines, just made Twitter # hashtag of contest and that's it.

I did need to ask from they that what are rules of this design competition, deadline, what kind of voice it has, what winner will get... Well, the answer: The deadline is 15 April 2020. No any samples yet because SiVo is still "work in progress" (yeah, it's not even yet in beta stage) and for design they hoped "cyber(futuristic)-like". For a prize, they have so far mentioned nothing...

Like *cough* seriously, it's really hard to me to believe in whole project to success until I can actually see/hear an actual alfa or beta stage program and has a Default voice for us to be heard. They have tweeted few times parts of "coding" but that doesn't much prove anything TBH. I also think that design contest has very ridiculous long deadline (2020?!??! 3 years??!?! Seriously???), it should be way more sooner!

I have become very doubtful of whole SiVo thing for now: they are good to talk but has nothing to show us - for now. I don't want to sound rude or discouraging but SiVo's developer should act more professionally in my opinion to get more seriously. For now, I just look at side how this gonna develop of will it develop at usable alfa/beta stage at all.

 

[Arissa]

About that "SiVo" thing that @Cheese mentioned:

To be really honest, I'm very skeptical of whole project at all. Ok, cool, the person behind it has realised that how VB6 support is ending in 2024 and because that they tweeted that "You cannot use UTAU anymore in 2024" - I remember, many MANY people in Twitter side get panicked!

Then they tweet that

"Because we cannot use UTAU in 2024 anymore, I introduce new program called SiVo that will replace UTAU. It's based on HMM coding."
"SiVo will support Japanese, English, Chinese, Korean *list continues* (...) for Japanese a talk voice too."
"I'm looking for female and male voicer for each language"

"To create a "voice bank" data for SiVo you will need to SING about 50 to 100 different songs" (yes, they told that VB will be made of actual REAL singing samples!) and them many people at Twitter have been like "but what if voicer is really bad at singing and cannot sing at all???" and "will UTAU voicebanks been supported in SiVo at all???" and after many this messages answer has been for that something like "will make support UTAU VB some way".

And just recently someone suggested that "what if you arrange a design contest for SiVo's coming "default voice"?" and of course they agreed and started it right away without giving contest rules or deadlines, just made Twitter # hashtag of contest and that's it.

I did need to ask from they that what are rules of this design competition, deadline, what kind of voice it has, what winner will get... Well, the answer: The deadline is 15 April 2020. No any samples yet because SiVo is still "work in progress" (yeah, it's not even yet in beta stage) and for design they hoped "cyber(futuristic)-like". For a prize, they have so far mentioned nothing...

Like *cough* seriously, it's really hard to me to believe in whole project to success until I can actually see/hear an actual alfa or beta stage program and has a Default voice for us to be heard. They have tweeted few times parts of "coding" but that doesn't much prove anything TBH. I also think that design contest has very ridiculous long deadline (2020?!??! 3 years??!?! Seriously???), it should be way more sooner!

I have become very doubtful of whole SiVo thing for now: they are good to talk but has nothing to show us - for now. I don't want to sound rude or discouraging but SiVo's developer should act more professionally in my opinion to get more seriously. For now, I just look at side how this gonna develop of will it develop at usable alfa/beta stage at all.

What you said makes sense. To me the whole thing does seem odd. It appears they've given little information and whatnot, have what looks like nothing they can really show us, and all in all it almost looks as though they're merely trying to replace utau...

But I'd have to look into it further myself.
[doublepost=1492111699][/doublepost]

There's been some talk in the Japanese fandom over a new vocal synth named SiVo. The developer made a blog post mentioning how VB6 support is ending in 2024 (the programming environment UTAU is written in) and how he's making this essentially due to that.

Many users have been tweeting about this and the possible end of UTAU once VB6 support expiresーand there's even a word for it nowーUTAU終了説 (UTAU Termination Theory).

Spoiler

Personally, I think the more vocal synths we have, the better. However, many people in the Japanese fandom aren't really taking to it being branded as an UTAU replacement and the whole panic that's been caused by this. I agree with that sentiment 100%.

semi off topicーbut I had actually emailed Ameya in regards to some things for the UTAU zine I'm organizing, and I was surprised to get a reply from him last night. Ameya still hasn't stated UTAU's development status, so if someone asked him he might reply...?

edit: I asked him about it in a followup email regarding zine stuff, so let's hope he'll reply.

So it wouldn't be possible for the UTAU program to be updated for it to continue working? I'm not very experienced in tech, but from what you mentioned about Sivo, would it be possible to keep UTAU if something was done regarding making it able to work? Like, doesn't Vocaloid and CeVio run on something, but I don't think Vocaloid will be inactive anytime soon, as it's quite well-managed, I'm sure.

 

[bio]

Ok so there's the Waltz editor and SiVo....
both im excited yet concerned for

 

[Kiyoteru]

From the photos I've seen, SiVo seems to be written in a programming language that's entirely Japanese. I'm worried about that.

 

[Nohkara]

I'm not expert with coding and doesn't know much anything about it but I know for sure that coding singing synthesiser from scratch is NOT piece of cake, you actually need real coding knowledge and experience: "knowing" isn't enough.

UTAU is build in VB6, to be supported in future for 100% sure I think/guess that UTAU needs to be re-build/re-coded from scratch in other thing. And to be honest... I think that creator of UTAU has no longer time and/or interests to do so.

Plus, I DON'T think that in 2024 UTAU will be immediately unusable: we have all ready now things like "virtual machine" and "wine wrapper" and I know that people can always think solution to run it way or another way.

IDK where/how UTAU-Synth is build but that MIGHT survive technically little longer than PC UTAU as long as we can still get working 30 day licence code in 2024 and later but I'm little sceptical of that too.

Sharpkey is very recent Chinese singing synthesiser (which has lot of potential TBH) and creator of Sharpkey has stayed once that they are working on that people can create own VB in that program but because the program is Chinese I think that it won't support no-Chinese VBs (at least not well) plus Windows and Chinese only for now.

I believe that if "making own virtual singer for free" idea reminds as a cool thing like now, someone will definitely make a "second generation free singing synthesiser" but this will take some time of course... Let's see what gonna happened.

 

[Tomato Hentai]

From the photos I've seen, SiVo seems to be written in a programming language that's entirely Japanese. I'm worried about that.

Maybe it's Dolittle?

 

[Nohkara]

From the photos I've seen, SiVo seems to be written in a programming language that's entirely Japanese. I'm worried about that.

Oh dear, that's NOT a good sight - specially when they have tweeted that "it will support multiple languages and lyrics input method". This makes me even more worried how that project gonna end up...

 

[KNΞMΛTCS]

The "singing songs" thing actually does make sense. As you mentioned, it is an HMM based synthesizer, not a raw sampler. HMM synthesizers are trained, not sampled. I'd bet it's using Festival or something of the like as the synthesizer. Although I wouldn't bet on it being done, seeing that it's just one fandom member. The only way we'll get a decent Utau replacement is either to have a very skilled programmer come and save the day, a lot of us to come together and try to build something, or a combination of both.

 

[Tomato Hentai]

Oh dear, that's NOT a good sight - specially when they have tweeted that "it will support multiple languages and lyrics input method". This makes me even more worried how that project gonna end up...

Just because the programming language is Japanese-based doesn't mean it'll be impossible to add-in an English-language GUI and methods of input for other languages. If it's programmed well, then it should be fine.

 

[Nohkara]

Just because the programming language is Japanese-based doesn't mean it'll be impossible to add-in an English-language GUI and methods of input for other languages. If it's programmed well, then it should be fine.

Ah, ok! Thanks for informing, I didn't know... ^^; (I hope that the program gets development well)

 

[Arissa]

The "singing songs" thing actually does make sense. As you mentioned, it is an HMM based synthesizer, not a raw sampler. HMM synthesizers are trained, not sampled. I'd bet it's using Festival or something of the like as the synthesizer. Although I wouldn't bet on it being done, seeing that it's just one fandom member. The only way we'll get a decent Utau replacement is either to have a very skilled programmer come and save the day, a lot of us to come together and try to build something, or a combination of both.

While singing songs may make sense, doesn't it still sound like a lot to sing 50 to 100 of them??

 

[na4a4a]

If it's HMM based then there goes my hopes tbh.

Voice training is out of reach for 99% of the current Utau community and also it never reaches the detail and tonal quality of a unit(sample) based voice...
Listen to Sinsy and Cevio, the quality isn't that great and male voice especially suffer from sounding like a blurry mess.
You need something around 4 hours of english recordings to do a decent HMM voice if I'm not mistaken. Depending on context this can be both more or less recording.
....though.... Japanese would take a lot less I wonder if you can ghetto-ly convert VCV utau voicebanks (which are usually somewhat redundant in vowels but not the transitions)...

And Festival by default is pretty pants at singing and so is everything else that's primarily a speech synth.
Also Festival by default is unit based but you can make voices that are HMM-like.
If anything they are probably doing something custom based on the available HMM/HTS source code.

---

Have we already forgotten about all the other Utau like projects though?
There are so many that it's hard to keep track.

https://twitter.com/_tori_san It appears this person is working on something.

And of course there is Masao's clone (which I cannot find a link for since I'm on mobile)


I don't get the hype for SiVo considering all the other projects that show just as much potential. It just happens to look aesthetically different from the others which made everyone go "oo shiiiny".

We need to stop looking at one project and then immediately dropping/forgetting about it when a new one comes along. That's crap behavior.

 

[KNΞMΛTCS]

I really don't care for the SiVo interface, actually. It just looks like the same old Utau editor with a fourth of your screen wasted to character art. And looking at it, it's probably using "eCantorix", part of Qtau development already. I highly recommend taking a peek at https://notabug.org/isengaara/qtau , as this fork is under active development.