You raise an interesting question about enforcement. Certainly that's going to be a challenge. Beginning with Napster the recording industry's approach was to target a select few to go after and send a message. I remember, many years ago, there were a lot of articles in the the news about some young woman, a teenager, that they went after because she had downloaded just a few songs. It was really kind of sad because the industry got a judgment against her for a few hundred thousand dollars which surely made a mess of her life and her future. Clearly, the idea was to send a message. It obviously didn't work since downloading never ended, but back then that was their approach.

The number of people downloading songs was in all likelihood much larger than the number of people experimenting with AI music composition, but it's still a lot of people. Going after all of them legally would be too costly and time consuming. They'll probably take the Napster approach and target a few to send a message of how much it will cost to proceed with what they're doing, and many, not all, will shut down out of fear the cost of continuing will be too great. I know as a fact the RIAA is already tracking many websites that are doing this sort of thing; scraping music recordings and then using emulations for AI music composition. At some point, as soon as any of them are really making money out of it, they'll pick some targets and start suing. I would guess between that, and their push for new laws that would prohibit some of this, will be their approach for a method(s) of enforcement.

 

Last edited: Apr 20, 2024

and everbody likes it when it swings...

 

I was thinking a file full of interesting quotes, actually. But then I noticed your glory wall. I have a text file I've been collecting for years, with quotes from various forums I found to be profound, interesting, or simply funny.

Like...
“Every record has been destroyed or falsified, every book rewritten, every picture has been repainted, every statue and street building has been renamed, every date has been altered. And the process is continuing day by day and minute by minute. History has stopped. Nothing exists except an endless present in which the Party is always right.”- George Orwell, "1984"

seems somewhat appropriate for the occasion...

 

Tried Udio recently. Pretty impressive what it can do! Created some pretty authentic sounding Philly Soul and Motown-style bangers with it.

I noticed however that in some situations it creates gibberish lyrics. It happens particulary if you use the extend function and for example add an intro to an existing song segment. The lyrics the AI wrote for the original segment (based on the prompt given by the user) are on topic, while the lyrics to the extended part are some messed up gibberish.

Also, what's the issue with the phasey and low quality sound of the results? Is this a deliberate limitation of the free beta version or due to the way the technology constructs music?


Example

the original segment created by my prompt has lyrics accurate to the prompt, but the intro I added through the extend functions has nonsense lyrics

original segment: https://www.udio.com/songs/c2uAb4AM3kyNA4rveZc74t
extended with intro: https://www.udio.com/songs/4bzHDutgA21Fr2RyqTNysM

Now I wish I could remove the singing in the intro part, because the intro is very dope instrumentally, but the nonsense lyrics make it hard to bear

The prompt was:
1960s soul & funk, philadelphia soul, strings, harmony vocal group, lyrics about feeling like an outsider in the cool crowd [manual mode NOT activated]

 

Last edited: Apr 23, 2024

Udio just released a new feature that allows you to upload your own audio and the model can add sections or extend it, you can add lyrics that will match the draft you're working on in terms of scale, mood and bpm, for example. It's a pity that you can't get the vocals on a separated stem as I know that they do the voice with a different model than the music and then they mix them toghether.


First 47 seconds are the draft I sent, the rest is spited by Udio based on some lyrics I put in there. This is slowly becoming a tool.



Disclaimer: I did 10 runs until this popped up. the rest where lyricless. Like all AI tools you have to try a lot to get some results.

 

Hey, finally Udio introduced stem downloads. So now you can upload you song, prompt, and take the results back to your DAW in a way that you can use them properly.