Voice cloning

The thing one has to consider is-

The statute affords damages to a person injured by another who uses the person’s “name, voice, signature, photograph or likeness, in any manner.”

There are a bunch of tort laws out there. And a bunch of already-won cases from this type of thing before current tech.

You plug in an ai-simulated Frank Sinatra voice to a released profit project…or even Morgan Freeman amongst hundreds/thousands of other known celebrity voice sonics…

let’s just say…there will be future litigation :slight_smile:

Personally, I would prefer to be on the sidelines rather than a target.

It’s one thing for a production to use respeecher in the context of their already-under-contract actors for the singular production.

It’s quite another thing for a piece of software to be marketed to the public where the end result of the use of the software is to provide an indistinguishable-from-the-real-thing Ol Blue Eyes sound.

In some regards, this is all going to become very messy.

Did you even read the use-case I listed as an example?

Essentially what you’re saying is that if you somehow change something so that it appears that Freeman said something he never said there could be legal issues. Well in post that happens almost all the time - through editing. And I doubt you’d object to DAWs allowing for us to change the order of the words Freeman spoke, right?

If anything I would say that any talent that values their voice as a signature “product” would actually be much more in favor of what I’m talking about than relying on crappy edits. Just imagine the difference between a video editor patching together what Freeman said into a new sentence, and then a mix engineer does the best they can but it sounds edited and choppy - that makes Freeman sound bad because of the edits. Compare that to re-reading the exact same sentence and running the ‘profile’ of his voice so that it is now the same recut sentence, but smooth. Which do you think he’d prefer? A crappy sounding Freeman, or a good sounding one?

It’s just a tool. Just like editing.

1 Like

No, that’s not what I’m saying.

I’m saying that if a guy in Miami…buys a commercial software program that can emulate Morgan Freeman’s voice…and the guy in Miami then constructs some sort of project from that software…and releases the Morgan Freeman soundalike work…and Morgan Freeman’s legal team works backwards…identifying that, yes, this commercial software that’s being sold at Guitar Center…is ai/deep learn machine code of Morgan’s voice…and Morgan didn’t authorize/contract for this software’s ability to do this…then there’s gonna be litigation down the road.

On the other hand, Morgan Freeman signs a contract to participate in a one-off project, his voice is manipulated during post in every which way…no problem. His contract stipulates that for the project as normal everyday stuff.

But…somebody asking Steinberg/Yamaha to release a commercial software program that can emulate every celebrity known to mankind…perhaps just by end-user whispering in some words and having them come out sounding like Dean Martin’s exact voice…well now…you’ve got some interesting issues with both estates and living celebrities.

As I see it

But litigation isn’t going to be successful. Software to do this already exists. Nobody as far as I know has been sued. How can the responsibility fall on Steinberg when a user uses the software illegally?

What’s the difference between that and me taking a bunch of movies and recutting Freeman’s voice into something he never said and making my own movie with his narration on it? Why would Steinberg be immune from a lawsuit in that instance but not when they release a feature that allows cloning the tonality of a voice?

It’s exactly the same problem: There are things we aren’t allowed to do as far as intellectual property goes. So far no DAW maker has been sued successfully - as far as I know at least - for any of the myriad of violations of IP that has happened over the years. For obvious reasons. Same reasons apply here.

Ask yourself this: How many (visual) deepfake companies have been sued so far?

Bonus question: What happens when Amazon, Microsoft, Google and Apple provide deepfake voice and image? Will they get sued successfully?

The risk here is near zero. Just put it in the user agreement that using the software legally is on the user, not Steinberg. Might already be in there in a generic sense… This is going to happen and anyone who’s asleep on this will be behind, goes for both engineers and DAW makers.

1 Like

Okay so, let me reiterate and explain again. This is already an ongoing issue, however the “fact checkers” is the direct solution to this. There are numerous photographs that have been photoshopped (for example “politicians photographed burning a bible, actor/singer sextapes, politicians giving the middle finger”) which have already been disproven by “fact checkers”. The difference between “respeecher” and “Adobe Photoshop” is that one is used for the visual aspects and the other is used for audio aspects.

Also to add, the goal for implementing a technology like this isnt for people to abuse (even though we already know that will most likely to be the case), the goal for implementing something like this should be for “content creators” to “create”. There are a lot of concept artist who use “Adobe Photoshop” to “create content” and they make a living doing so. That should be the intentions of implementing something like this.

1 Like

Voice cloning can be equated to this. Same moral questions arise. https://edition.cnn.com/2022/10/21/tech/artists-ai-images/index.html?fbclid=IwAR0IdYMnw3CCD-rM4RHTDDxV-FSvz4dFzVTB129fETOlWf9slNcTvMo3uhs

Visual AI can be prompted to make art in the style of any artist whose material has been used to train AI
…and nobody ask permission from owner to use intellectual property to train AI.

Any tool comes with responsibility how you use them. Even drill or hammer can be use very wrong…

Interesting article…

I want to add that I have been playing with “Unreal Engine 5” and the things I have been able to create is impressive to me. I’m by means no artist but because of the demos I have done, I believe I can be a vfx artist. Again I use that as an example to make the point that Spectralayers has a lot of potential but it’s up to Steinberg to decide if they want to heavily invest in it.

Mark Zuckerberg is going all out on his metaverse virtual reality and he spent billions on it and you could tell (just by how passionate he talks about it) that he is putting his all into it. You dont get that same enthusiasm from Steinberg when it comes to Spectralayers, it seems like Steinberg does little investments with the expectation of big rewards.

I would like to see a feature like voice cloning or respeecher within Spectralayers. It would be good if I could manipulate a voice or instrument in a unique way.

I also would like to see a feature like Pitch tracking within Spectralayers. In “RipX deep audio” you can not only isolate individual harmonics but you can literally move and change triplets/quartets of individual tones that are played in harmony/unison together. Celemony Melodyne comes close to doing something like moving the individual harmonics (triplets/quartets played in harmony) but it’s not smart enough to distinguish between sound timbres and most times it mangles voice with other instruments. It would be good if Spectralayers added Pitch tracking feature in a unique way where you can trace a sound(for example a trumpet) by it’s unique harmonic timbre profile. Sort of similar to the idea of using the harmonics tool but specifically geared towards individual sound sources. It would be interesting if I could not only manipulate a voice and change the characteristics but manipulate background vocals (for example singers who sing triplets/quartets in harmony unison together) and change the notes.

Might not be the best example as his company’s stock price tanked bad yesterday (after a long slightly slower decline) because the market thinks he’s wasting lots of resources on his pet project.

Also ‘passion’ is not a word most would associate with The Zuck.

The area marked in green is about when they announced the name change and focus to Meta.

If only Elon would takeover SpectraLayers, he’d make everything EZ…

To be honest with you, I believe the Metaverse idea is going to fail ONLY BECAUSE HIS NAME(MARK ZUCKERBERG") IS ATTACHED TO IT. The whole Augmented Reality/Virtual Reality is a good idea but because its from Mark Zuckerberg and people dont like him I believe anything he does will fail regardless. It seems like Mark Zuckerberg is unaware that people dislike him and dont trusts anything he does, and the way he talks about his ideas/products gives the impression he is blind to "The Real World and how people really feel about him(pun intended).

I used Mark Zuckerberg as an example to demonstrate that he’s investing into that Metaverse idea because he wants people to join and be apart of the Metaverse. Whereas Steinberg invests very little and expect a huge reward. For example "Metahuman is a good idea and the fact that you can create photo realistic images from people is a great idea and that great idea came from a lot of investments.

Well now… it’s not like Steinberg has the funds Meta has, right?

That’s true, however realistically speaking (from a Real world Reality perspective) I dont believe licensing respeecher technology or voice cloning technology is worth 2.8 billion dollars. Even though "Respeecher is a great idea, I myself wouldnt invest billions into purchasing technology like that for Spectralayers if I was in charge at Steinberg.

Also lets not forget that (although many of us arent a fan of Mark Zuckerberg) he made some very wise decisions by purchasing Instagram and Whatsapp 10 years ago. Whatsapp is not only thee most downloaded app, it is singlehandedly thee most widely used messaging app (beating Apple’s Imessage). The key is to make the right investments with a clear goal in mind.