This may seem like a strange question to ask, but when you think about it, content made using Synthesizer V AI voicebanks could count as AI art. It’s an AI program that generates vocals, which are in turn music. Music is a form of art. Therefore, usage of Synthesizer V’s AI voicebanks count as production of AI art.
Most people these days think AI art is an immoral technology that will destroy the livelihood of human artists, but I don’t see it that way. I think AI art is a very useful technology, as it could allow one to dream up ideas that they may have trouble putting down on paper. Synthesizer V is very similar in its function; it’s a program that creates vocals for users who can’t sing, don’t have the voice that they want for their song, or are outright voiceless.
While my opinion on AI art is generally positive, there are 2 uses for AI art that I see as negative; use of AI art to create fake news, and use of AI art to fake the work of a human artist. Synthesizer V could be used for these functions because their AI voicebanks are trained on the work of real voice actors and singers (Such as Yoshino Aoyama for Koharu Rikka, or Emma Rowley for Solaria). Using vocal synthesizers for talking is already a popular thing to do, so if someone wanted to, they could make someone like Emma Rowley say something they would otherwise not say. (Thankfully, as a far as I know, this has not happened yet.)
What do you think of this issue? Or… am I just spouting nonsense at this point?
In my opinion I don’t see Synth V as AI art - it does produce human like sounds but it does not create the lyrics (like ChatGP can) and it does not create the melody (notes) being sung - other AI programs exist that can create these but Synth V does not. To me Synth V is more like a studio musician / singer who is brought in to perform a work that another artist - hopefully a human has created. I am pretty sure that you would have no problem getting a US Copyright on a song you created on your own with Solaria singing. You created it not Solaria or Synth V. There are plenty of good musicians around - but good vocalist are hard to come by. Solaria can have a beautiful voice - but it does not happen straight out of the box in my opinion and experience. I think it takes a serious amount of time to get Solaria (only voice I have worked with) to sound really human - more effort than just doing re-takes in the software. Plus just like a human voice Solaria needs de-essing, compression and a host of other goodies applied during mastering to sound real. That’s my 2 cents - it maybe worth more or less?
Jack_O_Cherry already worded it pretty good, the following is my opinion.
I see Synth V as a smart sample library. Afaik it just recreates the frequency response of the input it was trained on instead of giving us a sort of omni- 10TB library.
Either way, the voice packs created and sold are all subject to license agreements/contracts with the voice provider. That is not happening with the majority of AI generated media right now and the reason why many people are upset, justifiably so.
The points above pretty much covered it, but on the specific point of restrictions, even the restrictions which are made to address the problematic uses of the technology are woefully ineffective.
At the moment the main “consequence” of using AI-generated images is that the company might not be able to claim copyright ownership of the image. The problem is that this doesn’t matter in practice; the company still owns the copyrights to the character design depicted, and the media/product the image is used in.
For example, why would Disney care if they don’t own the copyright to a generated image of Mickey Mouse when they own the entirety of “Mickey Mouse” as a character? There’s no tangible impact or risk, even for the companies with a history of aggressively enforcing their copyrights.
These sort of restrictions are even less applicable to SynthV’s implementation where the rendered vocals are not used in any sufficiently isolated manner.
And of course there’s the issue of the criteria for these restrictions being somewhat poorly considered in the first place. None of the “AI” technologies available are actually “Artificial Intelligence”.
The ability to generate derivatives based on machine analysis does not mean the computer is being “creative”, it’s just following a complex probability tree in a purely mathematical sense. The illusion of “intelligence” simply comes from the new methods of analyzing large datasets and traversing that probability tree in a performant manner.
When it comes down to it I suspect many of those who stand to profit will turn around and say “well these restrictions don’t apply to us because it’s not really AI, we just called it that for marketing purposes.”
I agree with much of what’s been said.
Personally, I don’t regard using Synth V any more AI-related than a piano or string preset I might use. I don’t even use the Instant mode. And it actually requires far more input than a human singer would.
But I can understand why people object to work that’s mostly AI generated. I wouldn’t be interested in AI generated lyrics or music. Take away the human and you take away the meaning, just IMO.
I agree with Northwave. A performance that could take 3 hours to record with a real singer takes me almost 20 hours with Synth V!! So yeah, the pug-in may be AI asisted but I do think that it is more of a tool like a drum plug-in that allows for more human expressiveness and gives the chance for non singer to express themselves…
Anyway, you will always be confronted to people that criticizes your art, or the way you do it… AI is a great tool, It is on its way and nothing can stop it now, so I think that not being interested in as an artist will inevitably becomes a drawback. And unfortunately, any tool in the hand of certain human being can becomes harmfull…
IMO, A synthesized voice is no different to a synthesised musical instrument sample/patch.