PG Music Home Forums General Discussion Off-Topic AI and the Future

David Cuny
https://www.pgmusic.com/forums/ubbthreads.php?ubb=showprofile&User=15953
Brought up an interesting point regarding the use of AI voices.

Elvis Act will become law starting July 1. The ELVIS Act "prohibits people from using AI to mimic a person’s voice without their permission, and violations can be criminally enforced as Class A misdemeanors. The law also authorizes civil action against any person who violates this law."

What's your thoughts on this?

question is ... How is this different from a Elvis impersonator singing on a record? Or the thousands of Tribute Bands who are touring the country making money on artists work many without permission from the artists.

How is this different from using Real Tracks instead of Real Musicians?

I assume the difference would be that AI would sample Elvis’s voice to create a song. Different than an impersonator or I suppose. 🤷‍♂️

Bud

It's probable (and logical) that voices will be treated similar to physical likenesses. You can't use a photo of Elvis without permission (there are exceptions) and so...you can't use his voice. Am not aware of the conditions impacting impersonators and tribute bands. But the BiaB artists have been paid, and if you have purchased BiaB, you have a license to use what the artists have contributed, royalty free.

But I also understand how tricky this is. If I purchase vocal lessons from Elvis (ha!) and end up singing similarly to him, that's probably ok. But if AI is fed the (purchased) Elvis catalog and produces a vocal facsimile...not ok? Hmmm.

All I know is...a LOT of lawyers are going to make bank on this topic.

Yes, Yes, and Yes.

But what if you sampled a voice from a very good Elvis Impersonator, with his permission? And especially if you paid for the samples in one way or another?

AI is going to be a legal playground before the rules get settled down.

I'm glad I make my living playing music to a live audience. That seems safe for at least a few more years.


Notes ♫

The issue with making regulations of any sort is that they are difficult and, in some cases, impossible to enforce. How will it be proven that AI was the tool that was used to create the voice?

There will be people who will do what ever they want regardless of the legality.
Whether the regulations are enforceable also depends on how wide spread the violations are.
It also depends on what the rewards are for breaking these laws.

These new laws may in fact be useful to protect the likes of Taylor Swift, perhaps. It is unlikely that they will ever affect personally the vast majority of musicians. Whose voice is so valuable that it is worth going to court over?

This attempt to control technology by laws and regulations will fail in the case of people who are not subject to the enforcement.

Of course, we will continue to make laws and regulations and not because they are for our protection but because that are a business enterprise in and of themselves.

This AI cat is already out of the bag and will take more than regulations to control it.

Just so you can read the actual law, here it is. If this law will stand up in court remaines to be seen.
It has been said by the government of Tennessee that it is not the intent for this law to effect cover bands.
The law itself is does not exclude that possibility and that is one of the possible untended consequences. Not everyone is in favor of this law as it is written. That fact will likely also wind up in court.


SENATE BILL 2096
By Johnson
HOUSE BILL 2091
By Lamberth
HB2091
011887
- 1 -
AN ACT to amend Tennessee Code Annotated, Title 39, Chapter 14, Part 1 and Title 47, relative to the protection of personal rights.
BE IT ENACTED BY THE GENERAL ASSEMBLY OF THE STATE OF TENNESSEE:
SECTION 1. Tennessee Code Annotated, Section 47-25-1101, is amended by deleting "Personal Rights Protection Act of 1984" and substituting "Ensuring Likeness, Voice, and Image Security Act of 2024".
SECTION 2. Tennessee Code Annotated, Section 47-25-1102(4), is amended by inserting "individual," before "firm,".
SECTION 3. Tennessee Code Annotated, Section 47-25-1102, is amended by adding the following as a new subdivision:
( ) "Voice" means a sound in a medium that is readily identifiable and attributable to a particular individual, regardless of whether the sound contains the actual voice or a simulation of the voice of the individual;
SECTION 4. Tennessee Code Annotated, Section 47-25-1103, is amended by deleting subsection (a) and substituting:
(a) Every individual has a property right in the use of that individual's name, photograph, voice, or likeness in any medium in any manner.
SECTION 5. Tennessee Code Annotated, Section 47-25-1104, is amended by deleting subdivision (b)(2) and substituting:
(2)
(A) The exclusive right to commercial exploitation of the property rights is terminated by proof of the non-use of the name, photograph, voice, or likeness of
- 2 - 011887
an individual for commercial purposes by an executor, assignee, heir, or devisee to the use for a period of two (2) years subsequent to the initial period of ten (10) years following the individual's death.
(B) For purposes of subdivision (b)(2)(A), "use" includes the commercial availability of a sound recording or audiovisual work in which the individual's name, photograph, voice, or likeness is readily identifiable.
SECTION 6. Tennessee Code Annotated, Section 47-25-1105, is amended by deleting subsection (a) and substituting:
(a)
(1) Any person who knowingly uses or infringes upon the use of an individual's name, photograph, voice, or likeness in any medium, in any manner directed to any person other than such individual, for purposes of advertising products, merchandise, goods, or services, or for purposes of fundraising, solicitation of donations, purchases of products, merchandise, goods, or services, without such individual's prior consent, or, in the case of a minor, the prior consent of such minor's parent or legal guardian, or in the case of a deceased individual, the consent of the executor or administrator, heirs, or devisees of such deceased individual, is liable to a civil action.
(2) A person is liable to a civil action if the person publishes, performs, distributes, transmits, or otherwise makes available to the public an individual's voice or likeness, with knowledge that use of the voice or likeness was not authorized by the individual or, in the case of a minor, the minor's parent or legal guardian, or in the case of a deceased individual, the executor or administrator, heirs, or devisees of such deceased individual.
- 3 - 011887
(3) A person is liable to a civil action if the person distributes, transmits, or otherwise makes available an algorithm, software, tool, or other technology, service, or device, the primary purpose or function of which is the production of an individual's photograph, voice, or likeness without authorization from the individual or, in the case of a minor, the minor's parent or legal guardian, or in the case of a deceased individual, the executor or administrator, heirs, or devisees of such deceased individual.
SECTION 7. Tennessee Code Annotated, Section 47-25-1106(a), is amended by deleting "photograph, or likeness" and substituting "photograph, voice, or likeness".
SECTION 8. Tennessee Code Annotated, Section 47-25-1106, is amended by inserting the following as a new subsection:
(f) Where a person has entered into a contract for an individual's exclusive personal services as a recording artist or an exclusive license to distribute sound recordings that capture an individual's audio performances, an action to enforce the rights set forth in this part may be brought by the person or the individual.
SECTION 9. Tennessee Code Annotated, Section 47-25-1107, is amended by deleting "photograph, or likeness" wherever it appears and substituting "photograph, voice, or likeness".
SECTION 10. Tennessee Code Annotated, Section 47-25-1107(a), is amended by deleting "account" and inserting "account, to the extent protected by the First Amendment to the United States Constitution".
SECTION 11. Tennessee Code Annotated, Section 47-25-1107(c), is amended by deleting "had knowledge of the unauthorized use" and substituting "had knowledge or reasonably should have known of the unauthorized use".
- 4 - 011887
SECTION 12. Tennessee Code Annotated, Section 39-14-152(g), is amended by deleting "Personal Rights Protection Act of 1984" and substituting "Ensuring Likeness, Voice, and Image Security Act of 2024".
SECTION 13. This act takes effect July 1, 2024, the public welfare requiring it.

Billy

This is an interesting discussion and I thank everybody for their contribution.

Can a voice be copyrighted? Or trademark?
If so does it ever expire? Will Elvis's voice at one time in the future become public domain?

Do we have to put disclaimers on our releases such as: AI : Elvis not really singing

The AI company that seems to be on the leading edge of celebrity voices is Lalals.com a German based company. They have 100s of celebrity voices and growing more each day.

The more I learn, the more I fear the future.

Personally (as if that matters to the legal system) I don't think you should be able to copyright a person's voice.

On the other hand, if you use it to pretend Elvis or whoever is singing the song, it's fraud, and that shouldn't be allowed.

I think if AI generates the vocals (or whatever) on the song, the vocal credit should simply go to the AI App.

And example would be if the song is called “Future Rock”, the credit would read “vocals by ____ using samples of Elvis Presley's voice” where _____ is replaced by the AI app.

A voice cannot be copyrighted, but it can be trademarked.
See: https://www.backstage.com/magazine/article/voiceover-usage-rights-explained-76226/

However, I suspect that AI voices will run afoul of copyright laws. Imagine that you had a recording Elvis singing a song. You then took the words from the recording, re-arranged them, pitch shifted and time stretched them, and released the result as a new song. That would be a copyright infringement. Even though you transformed the original song into a new song, you were still using the copyrighted performance of Elvis.

AI companies believe they can get around the copyright infringement by saying the because the AI "learns" the vocal style, they are not copying the original performance.

But how the data is stored and transformed is irrelevant. The AI has literally been trained to construct an audio signal that resembles Elvis' voice within a certain margin of error. The fact that that the neural network can create a credible imitation of Elvis demonstrates that audio data has been encoded into the network - even if it's difficult to point to exactly where it is, or how the network uses it.

The copyright of a performance will eventually expire, and the recording will be in the public domain.

However, a trademark will not expire if it is in continued use.

According to the ELVIS law, unless you have rights to the vocals, you won't be able to use them at all.

As I mentioned in my original post, that they are using celebrity's names and likenesses to promote their service. That is enough to run them afoul with trademark issues.

With the upcoming ELVIS Act, it will be illegal for companies to use unauthorized AI voices - at least in Tennessee.

A single note copied and used as a sample, even reversed, is a copyright infringement.

Ask those rappers who were successfully sued for doing that very thing a while back.

Why in Gods name would we need someting like this?

I hope they never release this. This would be a very dangerous tool for politicians! However I would have a ball with it

Thanks, --- food for thought... Drum machine > Autotune > Samples > BIAB > AI plugins > AI music.
Reply written by MS Copilot Chat GTP 3.5):
Video summary:

The video, featuring Rick Beato and titled "Did AI Just End Music? (Now it’s Personal)," explores the impact of AI on music creation. It discusses AI-generated music platforms like Udio and Sunno AI, which allow users to create music with simple text prompts. The video delves into the history of computer-assisted music, the technical workings of AI music apps, and the ethical concerns surrounding their use of copyrighted material. It also examines the potential effects on the music industry and musicians' livelihoods.

**Highlights**:
+ [00:00:00] **Introduction to AI-generated music**
* AI platforms can create complete songs from text prompts
* Examples of AI-generated music in various genres are showcased
+ [00:03:04 ]**Impact on the music industry**
* AI music systems could affect musicians and the industry
* Discussion with musicians Rick Beato and Taryn Southern
+ [00:10:09] **Technical workings of AI music apps**
* Explanation of neural networks and audio diffusion
* Comparison of AI music apps like Udio and Sunno AI
+ [00:14:00] **Ethical concerns and copyright issues**
* Debate over AI's use of copyrighted material
* Open letter from artists against AI infringement
+ [00:16:14] **Future of music with AI**
* Speculation on AI's role in music charts and creativity
* Personal reflections on the use of AI in music creation

Whether AI's impact on music is good or bad can be subjective and depends on various perspectives.

From a **creative standpoint**, AI can be seen as a powerful tool that enables creators to generate music in new and innovative ways. It can help musicians experiment with different styles and sounds, and even assist those without musical training to create their own compositions.

However, from an **ethical standpoint**, there are concerns. The use of AI in music raises questions about copyright infringement, especially if the AI is trained on existing copyrighted music. This could potentially harm artists who rely on royalties from their work.

From an **industry standpoint**, the rise of AI-generated music could disrupt the traditional music industry. If AI becomes capable of producing chart-topping hits, it could potentially affect the livelihoods of professional musicians.

In conclusion, like any technology, AI in music has its pros and cons. It's important to have ongoing discussions about how to best use this technology in a way that benefits everyone involved.

How will PGM embrace?
Here’s a sample response:
Written by MS Copilot Chat GTP 3.5):

“AI’s impact on music is indeed profound and transformative. As PGM Peter Gannon Music, we recognize the potential of AI as a tool for creativity and innovation. However, we also understand the ethical and copyright concerns that come with it. We believe it’s crucial to have clear guidelines and policies for AI-generated music to ensure fair use and protect the rights of all artists. The rise of AI in music presents both opportunities and challenges for musicians. While it could open up new avenues for creativity, it could also disrupt traditional structures in the music industry. Looking ahead, we see a future where AI and musicians coexist and collaborate, each bringing unique value to the music creation process. We are excited to be part of this journey and are committed to navigating these changes in a way that respects and uplifts the music community.”

As Peter would say... Have Fun!

There is probably already an app that does deep fakes for politicians. With samples to fake voice, too.

Notes, I think most politicians are deep fakes

Is this law only valid in the US?
How could you enforce it if the server producing the AI was on Christmas Island, say.

Can one country now legislate for the rest of the world?

I guess it's a copyright thing so if France for example didn't have this law, then writing a new Edith Piaf song would be ok.

Just my thoughts as usual but it sounds like the technology has shown existing laws to be out of date where AI is concerned.

Definitely fakes, but some are pretty shallow

Most of the legal stuff written here applies only to the US, though how much difference that actually makes I'm unsure.

One of the issues may well be where a song is created vs from where it's streamed vs where it's played. We here from time to time of cases being brought in various countries, simply because there's a better chance of winning and/or the payouts are likely to be better.

[*****] you, AI

It just learns by reading, watching and listening just like humans.
It is more prevalent than you think. Do you use a smartphone? Have you played with stem separation or chord recognition?
It is better to learn how to use it for your situation.

Certainly AI "learns".

However, AI does not learn "just like humans".

AI learns by looping through training data, slowly altering the weights in the neural networks to minimize the error value.

To recognize a single object, it typically needs thousands of examples. Training sets are massive, and typically accompanied by human-created tags.

Not the same thing at all.

Yes, I was speaking figuratively. People learn through repetition too.

The real questions to play out in courts is whether or not it has broken any copyright laws.

Experts in AI say no. People on internet are all over the board on the topic.

Most likely copyright laws will have to change.

I think it's dangerous to conflate the two as similar, any more than a horse is similar to a freight train.

Most likely. AI companies would love the laws to allow copyright to be assigned to AI created works.

Well, what's in the courts is the other way around but will be hard to prove when it creates at the pixel/bit level.

If you go listen to the whole video with Rick, he made a comment to the effect "when will you be a cover band for AI".

I think we will have AI "artist" who will become popular in the same way real artist are popular. If people like a song I don't think they will care if it is AI generated. If people like a video image I don't think they will care if it is real or not. This is really not something new. Snow White comes to mind....
I think we will have AI impersonaters the same way we have Elvis impersonaters.
I also think the advancements in AI are coming in the very near future, less that two or three years.

Imagine that you had never seen Michel Jackson in person. Consider if he never existed and today through techonology someone created his likeness and whole body of work. Would those music videos be any less interesting?

One thing that has been consistently true is that technology continues to advance no matter what laws we put in place to prevent it.
Even laws that address such serious issues such as nuclear proliferation are in general meaningless. Do you really think there is no condition that certain world actors would be stoped from pushing the button?



It's most likely not AI that we should fear, it is people. At least not in the near future.

Cheers,

Billy

+++ HERE +++ is an Associated Press article that details how artificial intelligence was used to create vocals for a new Randy Travis song.

Randy Travis is a an award winning country music singer who lost his voice to a stroke. +++ HERE +++ is a video of his new song.

This is so interesting. One of my YouTube subscribers told me about this a couple of days ago and I made a play along video. Nice song![video:youtube]Where that Came From (capo 4) by Randy Travis play along with scrolling guitar chords and lyrics [/video]

Nice song and vocals.

The idea of using AI for Randy's voice didn't come from his team, but from Warner Music Nashville Co-President Cris Lacy. My guess is that Warner Music thought the best way to introduce AI vocals was with a "feel good" story, and Randy was happy to oblige. In fact, that's exactly what they say in a press release:


From a technological perspective, the AI vocals something more akin to RVC, in that they need a "real" vocalist to sing the song. James Dupré was that vocalist. He was someone who had toured with Randy after his stroke, performing as Randy's voice. So they'd obviously had a good working relationship.

The recording used for Randy's song was one that James had done decade earlier, but was never released.

I initially thought this was cool since everyone involved was fully informed, but that's not the full story.

Rolling Stone wrote that James "was just as surprised as everyone when he heard Randy Travis’s AI-assisted vocal", noting that "Dupré was stunned when he realized that it was actually his voice he was hearing, at least in part."

James was paid for the rights to use his performance. But Rolling Stone writes that the Warner rep "declined to specify whether it was a one-time work-for-hire payment or if Dupré will participate in royalties."

And instead of giving James "sung by" credit, Warner created a new “vocal bed” tag that was coincidentally ready be rolled out just in time for the song.

I've got to wonder: Warner got a hold of an AI team in London, supplied them with vocal stems to 54 of his songs, got together musicians to create the backing for the song, and paid producers to spent months to work on the track.

How could they have forgotten to contact James about using his vocals?

My gut feeling is that Warner wouldn't spend all that money on the song if there was the slightest chance that James wouldn't allow them to use his vocals. I'd bet that Warner's lawyers had already concluded James' track had been "work for hire", and conveniently could be used however Warner wanted. This included use as the basis of an AI vocal - something that didn't even exist when it was recorded.

I would hope that James a bit of additional compensation, because who wants to have the a feel good story about the ethical use of AI ruined by the "vocal bed" singer? At the same time, I doubt that they wanted to set a precedent of "vocal bed" tracks being eligible for royalties. If they were, they wouldn't have evaded the question when asked.

While some news articles were quick to point out that the basis of the song was James' performance, it was notably absent elsewhere.

James isn't even mentioned on the YouTube video credits, and reading articles like the one in The Los Angeles Times, you'd never know any vocalist other than Randy was involved in the creation of the song.

What's problematic is that James' performance of the song wasn't good enough to release, but it was good enough to use as the "vocal bed" that was overlaid with Randy Travis' voice. Not really surprising, though.

All in all, a bit less of a "feel good" story for me.

I have never worked for Warner but having worked for both Polygram and Universal I can say all this does not surprise me.

What they are willing to spend money on defines any logic.

Billy