Fire your singers folks, Vocaloid 5 is here!

[wagtunes]

The original use for vocaloid in 2003 when it launched was to be used for temp tracks and experimental work and the like.

The things vocaloid has going for it is having a selection of virtual singers with different attributes and tones that vary from your own voice, the temp track options, and use as backup singers. If you combine a real voice with vocaloid the real voice will hide a lot of the flaws of the synthetic voices. Also don't need to bother with recording.

Don't think we're trying to replace a human singer, despite the thread title. There is a market both for the production and consumption of Vocaloid content, and that market is growing.

Well, the technology is not quite there yet, but I don't think we are far away from it. I hope it happens soon. We have digital simulations of Moog synths, Abbey Road plate reverbs and a lot of other great gear that very few of us would have a chance to play with some years ago. The advantages of getting really good voice technology could get us even further than real singers. Imagine beeing able to revive digital voices from the past like Elvis, George Michael, Jenny Lind or Edith Piaf and have it at your disposal to sing any tune in any style you like? Wouldn't that be cool? Listening to Google Duplex makes me think it might not be so far away.

Be careful when you say stuff like that because I actually started a thread a couple of months ago saying exactly this. I was almost lynched.

viewtopic.php?p=7033172

[Pelicanomicon]

It doesn't seems to be so easy to make voices for UTAU, but there is some people enjoying doing it, I guess that Vocaloid cant be much harder than UTAU so I guess some would do it. I have a five year old that love to sing so I could see myself spending a lot of time on it just cause it would be cool to have my kid as a Vocaloid. But from I have digged up about it in forums it seems that you have to have some "corporate license" for it. So it may not only be a matter of blood sweat and tears, but also a big wad of cash to pay Yamaha.

The company behind crowdfunded Vocaloid Tohoku Zunko asked for (and got) abt $50,000.

Other charas in the same mascot series as her were released as free UTAU banks
zunko.jp/con_voice.html (warning: anime!)

you can listen to the wav files inside the zips to get an idea of how singing robot sausage is made

The Japanese translations may be a bit off, but if I understand it this is just one voice bank for Vocaloid3, right?

The original crowdfund was for one Zunko VB, yeah. Later on she was reissued for Vocaloid4.

[PeterP_swe]

I'm optimistic that they will manage to model the human voice in the not too distant future, maybe 5-10 years.

I guess it will be though creating a CAD model of the actual throat cavity, animating through the vowels and other mouth key frames and then running some form of stochastic ray tracing for the air.

Maybe they'll "sample" singers with 3d scanners and motion capture.

...anyway, back to reality. Vocaloid 5 looks like fun and useful tool for scratch vocals.

[GearNostalgia]

The original use for vocaloid in 2003 when it launched was to be used for temp tracks and experimental work and the like.

The things vocaloid has going for it is having a selection of virtual singers with different attributes and tones that vary from your own voice, the temp track options, and use as backup singers. If you combine a real voice with vocaloid the real voice will hide a lot of the flaws of the synthetic voices. Also don't need to bother with recording.

Don't think we're trying to replace a human singer, despite the thread title. There is a market both for the production and consumption of Vocaloid content, and that market is growing.

Well, the technology is not quite there yet, but I don't think we are far away from it. I hope it happens soon. We have digital simulations of Moog synths, Abbey Road plate reverbs and a lot of other great gear that very few of us would have a chance to play with some years ago. The advantages of getting really good voice technology could get us even further than real singers. Imagine beeing able to revive digital voices from the past like Elvis, George Michael, Jenny Lind or Edith Piaf and have it at your disposal to sing any tune in any style you like? Wouldn't that be cool? Listening to Google Duplex makes me think it might not be so far away.

Be careful when you say stuff like that because I actually started a thread a couple of months ago saying exactly this. I was almost lynched.

viewtopic.php?p=7033172

Like you I don't get it why people are so conservative all the time. In the seventies the general consensus was that electronics and synthesizers was a bad thing for music and "dead" unreal instruments would make musicians unemployed. In the eighties analog synths were scorned as useless cause they did not sound real and could not sound like electric guitars and such. The most laughed at pieces was the TR808 that sounded like crap an no where near a real drumset and the totally useless shit TB303 that did not sound like a real bass. About at the same time there were ads and petitions to ban drum machines and sequencers cause they would make orchestras and drummers unemployed. Has any of this turned out in the slightest way as the negativists preached? Technology will move forward. It will give us new possibilities(and new threats). Roll with it, use it as best as you can, but trying to ignore it or stop it is as futile as trying stop the rain with your bare hands.

[wagtunes]

The original use for vocaloid in 2003 when it launched was to be used for temp tracks and experimental work and the like.

The things vocaloid has going for it is having a selection of virtual singers with different attributes and tones that vary from your own voice, the temp track options, and use as backup singers. If you combine a real voice with vocaloid the real voice will hide a lot of the flaws of the synthetic voices. Also don't need to bother with recording.

Don't think we're trying to replace a human singer, despite the thread title. There is a market both for the production and consumption of Vocaloid content, and that market is growing.

Well, the technology is not quite there yet, but I don't think we are far away from it. I hope it happens soon. We have digital simulations of Moog synths, Abbey Road plate reverbs and a lot of other great gear that very few of us would have a chance to play with some years ago. The advantages of getting really good voice technology could get us even further than real singers. Imagine beeing able to revive digital voices from the past like Elvis, George Michael, Jenny Lind or Edith Piaf and have it at your disposal to sing any tune in any style you like? Wouldn't that be cool? Listening to Google Duplex makes me think it might not be so far away.

Be careful when you say stuff like that because I actually started a thread a couple of months ago saying exactly this. I was almost lynched.

viewtopic.php?p=7033172

Like you I don't get it why people are so conservative all the time. In the seventies the general consensus was that electronics and synthesizers was a bad thing for music and "dead" unreal instruments would make musicians unemployed. In the eighties analog synths were scorned as useless cause they did not sound real and could not sound like electric guitars and such. The most laughed at pieces was the TR808 that sounded like crap an no where near a real drumset and the totally useless shit TB303 that did not sound like a real bass. About at the same time there were ads and petitions to ban drum machines and sequencers cause they would make orchestras and drummers unemployed. Has any of this turned out in the slightest way as the negativists preached? Technology will move forward. It will give us new possibilities(and new threats). Roll with it, use it as best as you can, but trying to ignore it or stop it is as futile as trying stop the rain with your bare hands.

Thank you. Somebody who gets it.

[ENV1]

I think the technology per-se existed in the early 80s, (SoftVoice), only it hasnt been developed (nor continued to be developed) with singing in mind. And thats a real shame because the potential is so obvious. Sure, one can use it just fine to make those robotic spoken bits and Kraftwerk did that a lot. But the engine is capable of so much more. Basically all thats needed is some more development to make everything a bit smoother, plus a real editor which lets you access the tons of internal parameters which the user usually never sees. Also since this thing is totally synthetic there would be no need for samples and stuff either, which means great flexibility. (Male, Female, Child, anything.)


Like a demo? Get it here.

Its a little rough around the edges when totally dry but for 1984 technology it certainly aint bad!

(At the very least it shows the potential, plus you can hear how well even such an ancient voice could work if you mask the rough stuff a bit.)


PS: The 'Audio Demo' button thats supposed to play the file directly doesnt seem to work.

Just download the WAV if it doesnt work for you either.

[Delta Sign]

We will certainly be able to synthesize vocals in the future, but controlling the synthesis engine will be the hard part.
There is a lot more going on than just volume and pitch, like a lot. It would probably take thousands of interacting parameters which all have to be automated in a meaningful way. Take a simple word like "is" and pronounce it in every way you can think of....yeah, see you in 100 years. The human voice is very complex and our ears and brains evolved to pick up and interpret all the tiny inflections.

Don't get me wrong, there will be improvements, but I don't think we will ever go beyond the "yeah, it sounds human" point without sacrificing flexibility. I love synthesized speech, but I'm quite certain it will never be able to replace real humans.

Just a random example of what I'm talking about: https://www.youtube.com/watch?v=4qw0PYkttk0

All those tiny inflections, pronunciations, dynamics and the crazy switching between modal and falsetto register he does (consciously or subconsciously) would be impossible to automate.

[GearNostalgia]

We will certainly be able to synthesize vocals in the future, but controlling the synthesis engine will be the hard part.
There is a lot more going on than just volume and pitch, like a lot. It would probably take thousands of interacting parameters which all have to be automated in a meaningful way. Take a simple word like "is" and pronounce it in every way you can think of....yeah, see you in 100 years. The human voice is very complex and our ears and brains evolved to pick up and interpret all the tiny inflections.

Don't get me wrong, there will be improvements, but I don't think we will ever go beyond the "yeah, it sounds human" point without sacrificing flexibility. I love synthesized speech, but I'm quite certain it will never be able to replace real humans.

Just a random example of what I'm talking about: https://www.youtube.com/watch?v=4qw0PYkttk0

All those tiny inflections, pronunciations, dynamics and the crazy switching between modal and falsetto register he does (consciously or subconsciously) would be impossible to automate.

It used to be the same arguments about the driftin in VCOs and the impossibilities of cloning a real analog filter characteristics - too complex, can't be done. It may seem impossible now, but advances in AI is going fast and when we have quantum computers I am sure it will just explode. We are getting there, and pretty soon is my guess. For good and bad.

https://www.youtube.com/watch?v=JvbHu_bVa_g
https://www.youtube.com/watch?v=MT_u9Rurrqg

[Delta Sign]

You are absolutely right about emulating the voice, but I'm not talking about the characteristics of the voice themselves, I'm talking about controlling them in a creative context. The google thingy automatically decides how to pronounce things based on what it has learned, in a creative context the user has to be able to make those decisions, and that's where things get hard.

[dreamkeeper]

... but advances in AI is going fast ...

Ah... you know, couple decades ago, when AI was still in its infancy, people where saying that already. In fact, the AI acolytes were singing that mantra so often, that a lot of people got huge angst, they could be losing their jobs or being enslaved by the machines. Well, some did lose their jobs - but certainly not because of AI.

Once the stage is reached where an AI singer could be considered as being good enough, you better be prepared, that (s)he'll say: "Eh... I don't feel like singing today. Please tell me a story instead! Or let's play some tic-tac-toe..." - which I personally believe, will be far in the future.

Oh, and the Vocaloid stuff sounds horrible - all of it!

[baggagelizard]

Next you all will be telling me you hate Delay Lama too

[Scotty]

Give it ten years ... you'll be eating your words.

... but advances in AI is going fast ...

Ah... you know, couple decades ago, when AI was still in its infancy, people where saying that already. In fact, the AI acolytes were singing that mantra so often, that a lot of people got huge angst, they could be losing their jobs or being enslaved by the machines. Well, some did lose their jobs - but certainly not because of AI.

Once the stage is reached where an AI singer could be considered as being good enough, you better be prepared, that (s)he'll say: "Eh... I don't feel like singing today. Please tell me a story instead! Or let's play some tic-tac-toe..." - which I personally believe, will be far in the future.

Oh, and the Vocaloid stuff sounds horrible - all of it!

[jancivil]

The original use for vocaloid in 2003 when it launched was to be used for temp tracks and experimental work and the like.

The things vocaloid has going for it is having a selection of virtual singers with different attributes and tones that vary from your own voice, the temp track options, and use as backup singers. If you combine a real voice with vocaloid the real voice will hide a lot of the flaws of the synthetic voices. Also don't need to bother with recording.

Don't think we're trying to replace a human singer, despite the thread title. There is a market both for the production and consumption of Vocaloid content, and that market is growing.

Well, the technology is not quite there yet, but I don't think we are far away from it. I hope it happens soon. We have digital simulations of Moog synths, Abbey Road plate reverbs and a lot of other great gear that very few of us would have a chance to play with some years ago. The advantages of getting really good voice technology could get us even further than real singers. Imagine beeing able to revive digital voices from the past like Elvis, George Michael, Jenny Lind or Edith Piaf and have it at your disposal to sing any tune in any style you like? Wouldn't that be cool? Listening to Google Duplex makes me think it might not be so far away.

Be careful when you say stuff like that because I actually started a thread a couple of months ago saying exactly this. I was almost lynched.

viewtopic.php?p=7033172

Like you I don't get it why people are so conservative all the time. In the seventies the general consensus was that electronics and synthesizers was a bad thing for music and "dead" unreal instruments would make musicians unemployed. In the eighties analog synths were scorned as useless cause they did not sound real and could not sound like electric guitars and such. The most laughed at pieces was the TR808 that sounded like crap an no where near a real drumset and the totally useless shit TB303 that did not sound like a real bass. About at the same time there were ads and petitions to ban drum machines and sequencers cause they would make orchestras and drummers unemployed. Has any of this turned out in the slightest way as the negativists preached? Technology will move forward. It will give us new possibilities(and new threats). Roll with it, use it as best as you can, but trying to ignore it or stop it is as futile as trying stop the rain with your bare hands.

Thank you. Somebody who gets it.

Which people? This is a link to a thread a while back where the issue of bringing people's wholly formed personalities to use as a virtual instrument was broached. I didn't like the idea, and would stand by everything I said there. It has nothing to do with synthesizers in any way. I was 11 or 12 when Wendy Carlos made Switched-On Bach. Which has nothing to do with replacing any instrument. I was using a Minimoog by 1974. It was a synthesizer. I was not replacing anything by use of it. I bought a drum machine when that was fairly new. I was actually acting as the drummer in my bands behind it. Now, while some people would see us and hear drums and 'where's the drummer' instead of me sitting behind a table, I was not replacing any drummer except for not having to be arsed with one. I was under no illusion.

Who made these arguments? Do y'all not know what a straw man is? I do remember those sentiments, by people with a vested interest and some fear, and in particular the Musician's Union was behind some of it. I thought it was stupid even as a child. I have nothing in common with that kind of thinking. Who does? Who did that?

I'm not "conservative". The word you want is more like 'reactionary'.
I have nothing to do with those views. I have not seen these views here at KVR, even.
I'm a forward-thinking composer (unlike some, frankly; "I don't get all this atonal music."). I think, and I said so, that capturing an extant whole identity in such an instrument seems unethical to me. In a deeper consideration than some people were interested to deal with, it looked like to me. I think there are ramifications to it. I will reiterate I don't think it's necessarily possible to this extent but as what we call a thought experiment it is far from an appealing notion to me. My arguments are there and beyond this cursory mention of it I don't think I'll reconstruct all of it here. It does resemble a bit what I did just see -

Once the stage is reached where an AI singer could be considered as being good enough, you better be prepared, that (s)he'll say: "Eh... I don't feel like singing today. Please tell me a story instead! Or let's play some tic-tac-toe..."[...]

*

- in that thought experiment sort of way. I personally doubt that the species will be around long enough for it.
I think Philip K Dick went into <what is human> and the idea of replication of it (a whole personality means a whole, an extensive and deep set of memories, which are believed by the replicant to be experiences) better than anything I could write... and the ethical consideration is at least implied.

This would literally be replacing something with tech. This Vocaloid is obviously a synthetic voice, and no one with all their faculties, who knows from hearing a human voice believes otherwise. Personally I would stay away from hearing it because the effect is not very musical as far as I'm concerned. There are guitar tones which I don't care for, for that matter (and I'm not a big fan of synth drums). My ear is my ear, it's probably more sensitive than some in certain ranges and besides that, this is that thing we call 'taste'. And aesthetics. I want something more than synth drums does, even as I understand the preference and viability for a certain style area. I'm very into the human voice...

Nowadays I am replacing musicians (except_when_I'm_not), and certain drummers I look up to are selling MIDI grooves. So the fears of the Musician's Union were borne out? Not really, because there were job opportunities created. I wonder how many of the users of Miroslav Philharmonic for Sampletank know who Miroslav Vituous is?

But one of the larger ramifications is dealt with in science fiction. In the broader sense, *'use it or lose it', confer The Time Machine by HG Wells.

[GearNostalgia]

Which people? This is a link to a thread a while back where the issue of bringing people's wholly formed personalities to use as a virtual instrument was broached. I didn't like the idea, and would stand by everything I said there. It has nothing to do with synthesizers in any way. I was 11 or 12 when Wendy Carlos made Switched-On Bach. Which has nothing to do with replacing any instrument. I was using a Minimoog by 1974. It was a synthesizer. I was not replacing anything by use of it.

Sounds like a deep thread. Please post it if you find it. I did not hit it with my random search.

There has been many different aproaches to using synths. I don't remember seeing any polls about how many wanted to use them to replace other instruments or band member, they may not have been the majority. But like you I also recall musicians unions were active advocating bans. Personally I was working part time in a music store in the 1990ies and there was not a single day when I did not get comments from customers that they did not want to buy synths cause the did not sound like the real instruments and they wanted to buy something that could replace other instruments. A lot of people wanted that so they scorned old "crap" like MiniMoogs, MonoPolys and Jupiters and bought stuff like JV80 instead cause they sounded real.

A lot of people may want to replace a real singer with an AI version. I would like to do that myself, but I also like Vocoders and I think Vocaloids have a certain charm in its "anime" sound that can work well in certain genres as well.

But I think it is like the splitting of the atom, if it can be done it will be done for better or for worse and I think AI singers are coming pretty soon.

[GearNostalgia]

Too bad it does not seem like they have opened it up for people to create there own vocies like with UTAU.

Actually you can. There are lots of 3rd party voice banks for Vocaloid not made by Yamaha. How you go about doing it? That's an entirely different matter. I do know this much. It's a crap ton of work. Trust me, you DON'T want to do it.

That's correct. But even if you could get a license and software for free, the process is not for the feint of heart. For one thing, you have to find a vocalist and record all the necessary phonemes, whatever they are. There are tons of them. Then you have to go through the conversion process and "fix" all the problem ones.

A typical voice bank takes a good year to do.

Do you REALLY want to go through all that? If you're that dedicated, a license and some software isn't going to stop you. So that part of it is a non issue.

This isn't something a casual user is going to do. Hell, I wouldn't even want to do it even if I could get Celine Dionne down here to record all the parts.

People do this for the purpose of selling the libraries and making money.

I wanted to find out more about how they license Vocaloid for developers and what such a license cost. yamaha just replied that they only issue licenses to other companies and gave no answer about how much such a license costs. So case closed. It can not be done by regular customers.