Hacker News new | ask | show | jobs
How I see the music industry of the future
2 points by matthodan 698 days ago
The other day I made a song on SUNO (suno.com/), a generative AI tool and holy moly did it deliver... In about 30 minutes, I had created an alternative pop song that is almost indistinguishable from a live recording. As I was making the song, I felt like I (a guy with no musical ability) was working with one of the most talented (and patient) bands to bring my song to life through brute force trial and error. This experience has singularly changed how I see the future of the music industry.

In the future, I don't think musical artists will make songs. I think they will make models. Those models will contain all of the information needed for generative AI tools to quickly and accurately create music that reflects the sound and compositional qualities that define that artist. AI tools will create new original music spontaneously that is customized to the listener, perhaps taking into consideration the listener's mood and activities. Net net, the principal unit of economic value in the music industry will shift from being songs, to models.

It is a brave new world.

7 comments

AI generated music is not protected by copyright, at least in the US, since copyright can only affix to works created by humans (the "human authorship requirement").[1][2] Without copyright protection, there is no "economic value" -- at least directly tied to the songs themselves. AI generated songs are not paid royalties at e.g. The Mechanical Licensing Collective for US streaming use.[3]

[1] https://www.govinfo.gov/content/pkg/FR-2023-03-16/pdf/2023-0... (Copyright Registration Guidance section)

[2] https://ipwatchdog.com/2022/02/23/thaler-loses-ai-authorship...

[3] https://www.copyright.gov/ai/USCO-Guidance-Letter-to-The-MLC...

There will still be live musicians, nothing will change that.

But I see it becoming like how people onstage use synths and other electronic gear to make music, maybe they won't use AI exactly realtime, but a mix of real instruments and AI. Also Rap started off using old records for the backtrack, then samples, so its probably going to be similar as that.

But there will be a lot of new music, how anyone finds the best stuff, I have no clue. I don't even listen to USA music or the radio, there is no need for over 20yrs thanks to the internet. There is a ton of stuff out there, and that will only grow.

But some think sites like Spotify and Pandora will use AI generated music so they don't have to pay royalties. It makes sense for study, instrumental type music. AI music is pretty impressive, but the singing is too static to me, it sings too many lyrics too fast and with the same intensity 8 times in a row. Of course it will get better, but it really needs to understand what words really matter.

And if you want to listen to my creation, here it is: https://soundcloud.com/the-squeeze-972740152/crushed
What happens to live music in this scenario?

How will new artists break into this market? Existing artists who are popular may be able to make these "models" as you call them, but how will new artists break into this market? How will they rise above the inevitable noise?

Actually, I see the opposite happening. I think we're going to witness the resurgence of the singer/songwriter and artists like Taylor Swift are already paving the way for that to happen. EDM might get off on what you're envisioning, but I don't think others are headed that way.

All great questions. With these generative AI tools, the effort required for making quality music has dramatically decreased. The big challenge becomes how do you stand out in a very crowded market. To break out, I think you almost need to have other artists (or influencers) who have already made it validate your work and expose you to their audiences.
I think AI can make contrived records that sound like every other record. I don't think it can make true timeless and classic music.

On a completely different note, I think the premise of this idea sounds absolutely dystopian.

I think this is true today, but timeless and classic music comes from creativity. And creativity is a trial and error process of building on prior approaches and techniques. Randomly testing new approaches and techniques by blending and extending past approaches and techniques seems like something generative AI (with human prompting and targeted supplementation) would be particularly good at.
It's not like the AI made this on its own, it was iterative work with a human

> As I was making the song, I felt like I (a guy with no musical ability) was working with one of the most talented (and patient) bands to bring my song to life through brute force trial and error.

Max Martin comes to mind as an interesting comparison

(https://en.wikipedia.org/wiki/Max_Martin)

No. Because that’s not why real musicians make music.

You’re describing a game you played. You’re not an artist.

Didn't I see this exact same post, shilling the exact same AI music app, earlier this week?
80 times in the past 12 months, almost twice a week:

https://hn.algolia.com/?dateRange=pastYear&page=0&prefix=fal...