The Biggest Problem With EVERY Vocal Creation Program

[wagtunes]

All I know is with SoundID I sing the song. I then turn on analyze and SoundID then replaces my vocal with the singer I chose.

What, with the exact same intonation, timing and accent?

Hmmm, makes sense... they absolutely definitely do the extra 100,000% percent work to 'replace' the original with entirely new generated audio, rather than just processing the original audio.
And they refer to it as doing 'transformation' and 'morphing' because words like 'creation' or 'generation' just arent AI enough.
Yeah, that makes sense.

But hey, you dont want to even try a suggestion because something something AI. So it wasnt really about solving anything, just whinging.

It's odd that this is such a contentious subject for people.

hmmm, irony.

But mention trying to get variety out of a vocal plugin and everybody has a cow.

I don't get the utter disdain for vocal programs.

well, just consider how you're reacting to the notion of an AI algorithm, and apply that.

Vocaloid, on the other hand, you draw notes and pick the singer you want to sing those notes.

I just want to know why I can't get voices that have a little more character to them

Because what Vortifex already told you.
Because the companies are predominantly japanese, and aim at a market that isnt you.

That makes sense. It explains why everything sounds like an anime production. Oh well, it is what it is.

[sonidomolino]

op is a good question, seems the distinction between "AI" (generative) and "AI" (transformative) gets lost in all the marketing/drama of AI in general.

i've tried Soundworks SoundID VocalAI of the "transformative" class, like an amp sim to change your EG DI input into a Marshall stack on 10. It does not generate melodies/character/emotion (i.e. a vocal performance).

While it might be adequate for a demo/guide track, issues were noted with phrasing, timing, attack, mangled vowels/consonants, etc. Could try digging deeper, not sure it's worth the effort at this stage. And for sure it gets replaced by real people when it counts.

per this;

You won't find ... anything with any real grit and character. So here's my question. Why?

... model(ing) the reflexive, fluid-structure interactions that a human singer uses to convey emotion.

Current technical limitations ... include:
1. The "Source-Filter" Oversimplification etc. etc. etc.

"character" = formant? is a function of the immensely rich number of physical attributes of an individual human. as noted, the challenge is in the complexity of modelling your specific formant and then transforming it to a target formant. all sounds pretty complex and computationally intense to me. the products i've seen come with a few options for "character", not clear if the op is looking to sound like someone specific or more general case.

the AI answer leans in the wrong direction by mentioning "emotion", which is a "performance". if you can't sing, the issue is talent, not a plugin, so maybe work on something you can do. jmho.

[kraster]

Now you get it.

So since it is ALREADY happening why not at least give us some voices that don't sound the same? There is no reason not to unless there is simply no demand for it and thus no money to be made from it.

And I thought you said you were out of this discussion.

I think the point is that once you start making a voice a "type" it is no longer unique.

What sets apart people like Rod Stewart, Joe Cocker, Neil Young, Dylan etc. is that their voices have a lot of nuance and individualism.

AI voices tend to flatten individualism by nature. There are voices in programs like SoundID that could be described as gravelly or nasal or whatever but they still conform to some kind of overarching ruleset.

People like Dylan or Joe Cocker for example have so many inflection and "imperfections" and specific character traits that they don't lend themselves to be tied up neatly in a model

[bermudagold]

Now you get it.

So since it is ALREADY happening why not at least give us some voices that don't sound the same? There is no reason not to unless there is simply no demand for it and thus no money to be made from it.

And I thought you said you were out of this discussion.

I think the point is that once you start making a voice a "type" it is no longer unique.

What sets apart people like Rod Stewart, Joe Cocker, Neil Young, Dylan etc. is that their voices have a lot of nuance and individualism.

AI voices tend to flatten individualism by nature. There are voices in programs like SoundID that could be described as gravelly or nasal or whatever but they still conform to some kind of overarching ruleset.

People like Dylan or Joe Cocker for example have so many inflection and "imperfections" and specific character traits that they don't lend themselves to be tied up neatly in a model

exactly...if it is that unique, it wouldn't be selected for by statistical inference in AI...but his fundamental premise is flawed anyway...that synthV is the only option for people who cant sing...all the AI vocal apps dont need you to be able to sing...they have pitch correction, they take midi, they allow you to insert breaths and timbre changes temporally...and they all take creative license with what you sing in to them anyway, as a forcing function towards a statistical mean of what they have been told through training as pleasing or good...I'm sure Neil, Aaron, Spencer, and Larry below are perceived to have some modicum of "character" and/or "grit" to the 95th percentile...which is all these tools are trying to achieve
https://www.ikmultimedia.com/products/r ... p?p=models

[wagtunes]

Now you get it.

So since it is ALREADY happening why not at least give us some voices that don't sound the same? There is no reason not to unless there is simply no demand for it and thus no money to be made from it.

And I thought you said you were out of this discussion.

I think the point is that once you start making a voice a "type" it is no longer unique.

What sets apart people like Rod Stewart, Joe Cocker, Neil Young, Dylan etc. is that their voices have a lot of nuance and individualism.

AI voices tend to flatten individualism by nature. There are voices in programs like SoundID that could be described as gravelly or nasal or whatever but they still conform to some kind of overarching ruleset.

People like Dylan or Joe Cocker for example have so many inflection and "imperfections" and specific character traits that they don't lend themselves to be tied up neatly in a model

exactly...if it is that unique, it wouldn't be selected for by statistical inference in AI...but his fundamental premise is flawed anyway...that synthV is the only option for people who cant sing...all the AI vocal apps dont need you to be able to sing...they have pitch correction, they take midi, they allow you to insert breaths and timbre changes temporally...and they all take creative license with what you sing in to them anyway, as a forcing function towards a statistical mean of what they have been told through training as pleasing or good...I'm sure Neil, Aaron, Spencer, and Larry below are perceived to have some modicum of "character" and/or "grit" to the 95th percentile...which is all these tools are trying to achieve
https://www.ikmultimedia.com/products/r ... p?p=models

Okay, I'm going to try to convey this as clearly and plainly as I can.

As a songwriter and NOT a performer, my goal is to get my song in front of somebody who likes it enough to want to record it. Now, one thing I have learned from professionals in this business is that the best chance you have to accomplish this is to make a demo that best conveys what you want your song to get across.

Let's take a very specific example. As you probably know, Barry Manilow did not write "I Write The Songs". The song was written by Bruce Johnston of the Beach Boys. Now let's say I wrote the song and I specifically wrote it for Barry Manilow. My best chance of getting him to do it is to give him a demo with a singer that is close in style to his. Now, I can't sing for crap. But with a vocal program that can emulate his voice, he can hear, at least pretty closely, what it would sound like if he did the song.

That is my goal as a songwriter. To produce demos that give my songs the best chance of being covered because i am no performer and freely admit that.

So, with a wide variety of voices, I can do MORE songs in MORE different styles. That in turn gives me MORE chances to get my songs covered.

I can't make it any clearer than that.

[zerocrossing]

I think it's all about the perceived market. They're chasing the young influencer set, which is stupid, IMO, because it's the old geezers that have fat cash, yo. I also think it's the older dudes like McCartney who are leaving money on the table. First off, his voice is shot. I bet he could squeeze out some more decent tracks if he had an AI voice trained on all his music to use for the notes he can't hit any more, and to lesser artists who could use the funds, set up a royalty deal and get a cut off of each sale. But if you can't get an artist, just train it with a tribute band singer. Better yet, just find some interesting singers that are looking for money and use them. The world is littered with great musicians who have never made a commercial success.

[zerocrossing]

I've already answered that. You just need to understand it.

So you're saying companies are purposely not making voices with character because they don't want to step on the toes of singers like Joe Cocker? Tell me you're kidding? Do you really think these companies care if a singer gets pissed off because they made a vocal plugin with character?

You just don't get it. What good would it do if a tool like that could mimic a certain “rough” voice, for example? Suddenly there would be 20 songs with, say, the same “rough” voice, or all with the same character. Is that what you want? Because that's exactly what would happen.

I'm out of this discussion.

I thought a lot about this, and came to the conclusion that, while it's not optimal, it's really no different than the millions of tracks that were all made with a Fender Stratocaster into a Marshall, or a Moog Model D, Roland Jupiter 8, etc.

I've said this before, but no one cares. We all think we're unique snowflakes, and maybe we are, but we're buried deep under permafrost so no one is ever going to experience us. They let a few drift out every now and then, to keep the illusion that there's a chance to "make it," but that's just so they can sell us stuff. I decided that I just wanted decent vocals and that could mean anything. My preference will be for someone to say, "hey, I'd like to collaborate with you and sing your track," but I've yet to get that offer, so Synthesizer V it is.

[bermudagold]

Now you get it.

So since it is ALREADY happening why not at least give us some voices that don't sound the same? There is no reason not to unless there is simply no demand for it and thus no money to be made from it.

And I thought you said you were out of this discussion.

I think the point is that once you start making a voice a "type" it is no longer unique.

What sets apart people like Rod Stewart, Joe Cocker, Neil Young, Dylan etc. is that their voices have a lot of nuance and individualism.

AI voices tend to flatten individualism by nature. There are voices in programs like SoundID that could be described as gravelly or nasal or whatever but they still conform to some kind of overarching ruleset.

People like Dylan or Joe Cocker for example have so many inflection and "imperfections" and specific character traits that they don't lend themselves to be tied up neatly in a model

exactly...if it is that unique, it wouldn't be selected for by statistical inference in AI...but his fundamental premise is flawed anyway...that synthV is the only option for people who cant sing...all the AI vocal apps dont need you to be able to sing...they have pitch correction, they take midi, they allow you to insert breaths and timbre changes temporally...and they all take creative license with what you sing in to them anyway, as a forcing function towards a statistical mean of what they have been told through training as pleasing or good...I'm sure Neil, Aaron, Spencer, and Larry below are perceived to have some modicum of "character" and/or "grit" to the 95th percentile...which is all these tools are trying to achieve
https://www.ikmultimedia.com/products/r ... p?p=models

So, with a wide variety of voices, I can do MORE songs in MORE different styles. That in turn gives me MORE chances to get my songs covered.

I can't make it any clearer than that.

That was never unclear to me...my only point is there are already quite a range of styles...did you listen to the examples I posted?...these should be more than good enough to convey the style and sentiment of your songs in a sketch

[bermudagold]

I think it's all about the perceived market. They're chasing the young influencer set, which is stupid, IMO, because it's the old geezers that have fat cash, yo. I also think it's the older dudes like McCartney who are leaving money on the table. First off, his voice is shot. I bet he could squeeze out some more decent tracks if he had an AI voice trained on all his music to use for the notes he can't hit any more, and to lesser artists who could use the funds, set up a royalty deal and get a cut off of each sale. But if you can't get an artist, just train it with a tribute band singer. Better yet, just find some interesting singers that are looking for money and use them. The world is littered with great musicians who have never made a commercial success.

but that is happening...all the big boys allow you to train ur own vocal models now...resing even lets you do it locally...and nothing is stopping you from cloning a famous persons voice...its just in the TOS you agree to not to do that, so if you get caught ur liable...star producers like jimmy jam for example believe that once a 2 step model is ironed out...permission and compensation...most legacy stars and their estates will be all for it...he points out that twice he has been stopped by artists on an airplane and at McDonalds to thank him for sampling them for Janet Jackson records because the publishing checks on platinum records bought one guy a house and another a new swimming pool...but permission to sample a song is different from permission to clone their voice and pump out hundreds of songs...it will likely come down to whether the artist or their estate needs the money...look how many huge artists have sold their publishing catalogues in the last decade...especially ones who had bad deals and don't own much publishing...salt and pepa catalog has brought in on avg 10 million usd a year since the 80s...and they haven't seen virtually any of that...im sure they would gladly sell their voices to cloning

[wagtunes]

Now you get it.

So since it is ALREADY happening why not at least give us some voices that don't sound the same? There is no reason not to unless there is simply no demand for it and thus no money to be made from it.

And I thought you said you were out of this discussion.

I think the point is that once you start making a voice a "type" it is no longer unique.

What sets apart people like Rod Stewart, Joe Cocker, Neil Young, Dylan etc. is that their voices have a lot of nuance and individualism.

AI voices tend to flatten individualism by nature. There are voices in programs like SoundID that could be described as gravelly or nasal or whatever but they still conform to some kind of overarching ruleset.

People like Dylan or Joe Cocker for example have so many inflection and "imperfections" and specific character traits that they don't lend themselves to be tied up neatly in a model

exactly...if it is that unique, it wouldn't be selected for by statistical inference in AI...but his fundamental premise is flawed anyway...that synthV is the only option for people who cant sing...all the AI vocal apps dont need you to be able to sing...they have pitch correction, they take midi, they allow you to insert breaths and timbre changes temporally...and they all take creative license with what you sing in to them anyway, as a forcing function towards a statistical mean of what they have been told through training as pleasing or good...I'm sure Neil, Aaron, Spencer, and Larry below are perceived to have some modicum of "character" and/or "grit" to the 95th percentile...which is all these tools are trying to achieve
https://www.ikmultimedia.com/products/r ... p?p=models

So, with a wide variety of voices, I can do MORE songs in MORE different styles. That in turn gives me MORE chances to get my songs covered.

I can't make it any clearer than that.

That was never unclear to me...my only point is there are already quite a range of styles...did you listen to the examples I posted?...these should be more than good enough to convey the style and sentiment of your songs in a sketch

By that logic. why do people need 100 vsts? But if they do, that's okay. Wanting more voices isn't?

Again, the hypocrisy and double standard is astounding.

[whyterabbyt]

By that logic. why do people need 100 vsts? But if they do, that's okay. Wanting more voices isn't?

Who said wanting it wasnt okay? The people that took time out to explain stuff to you or offer suggestions, only for you to spit 'you dont understand what I want, why cant I get it, I want this thing ' over and over again back in their faces.

Again, the hypocrisy and double standard is astounding.

Not as much as your childish ranting and umbrage. You're asking a random crowd why what you want doesnt exists, and people are people are telling you why, and offering alternative ways of achieveing what yu say you want. And you're being a f**king child about it. You're not somehow entitled to it existing just because you want it.
And the people that are trying to help you with your 'but I want it and you dont understand because me me me me me so why' arent the ones who somehow prevented it from being possible.
Grow the f**k up, seriously.

[whyterabbyt]

I forgot how WT 'but why' threads go, I guess other people do as well, otherwise noone would waste their time on them.

Victims' talent for high drama draws people to them like moths to a flame. Their permanent dire state brings out the altruistic motives in others. It is difficult to ignore constant cries for help. In most instances, however, the help given is of short duration. And like moths in a flame, helpers quickly get burned; nothing seems to work to alleviate the victims' miserable situation; there is no movement for the better. Any efforts rescuers make are ignored, belittled, or met with hostility. No wonder that the rescuers become increasingly frustrated – and walk away.

[wagtunes]

By that logic. why do people need 100 vsts? But if they do, that's okay. Wanting more voices isn't?

Who said wanting it wasnt okay? The people that took time out to explain stuff to you or offer suggestions, only for you to spit 'you dont understand what I want, why cant I get it, I want this thing ' over and over again back in their faces.

Again, the hypocrisy and double standard is astounding.

Not as much as your childish ranting and umbrage. You're asking a random crowd why what you want doesnt exists, and people are people are telling you why, and offering alternative ways of achieveing what yu say you want. And you're being a f**king child about it. You're not somehow entitled to it existing just because you want it.
And the people that are trying to help you with your 'but I want it and you dont understand because me me me me me so why' arent the ones who somehow prevented it from being possible.
Grow the f**k up, seriously.

I understand and accept why what I want doesn't exist. There is no market for it. But telling me that I don't need it because I have enough choices already, that's what I take exception to. If people can have 7 wavetable synths then there is nothing wrong with me wanting more voices regardless of whether or it's ever going to happen.

Maybe YOU'RE the one who need to grow the f**k up with your constant insults?