15 Comments
User's avatar
Monkeyb00y's avatar

Suno v5 came out today, Sept 23.. Tomorrow (Sept 24th) Studio.

Can't wait to try it out.

Vick's avatar

The second. It sounds more natural and not trying to be too fast where the first just sounds rushed.

GH's avatar

I like the second song better as it sounds more "open".

I redid all my music in Suno recently too, and it's really fun. Very few musicians can do what it spits out in seconds, and I can just get another version in if something didn't work out.

It's the first slot machine I enjoy pulling.

Monkeyb00y's avatar

Great tune. Reminds me of Yellowcard minus the violin.

Sean Patterson's avatar

2nd. Took me a while tho'.

There was a metallic sound in the first.

Mark Pierce's avatar

U.S. registrations require disclosure of the AI‑generated material and a description of the human contribution. That means every claim starts narrower. If the registered copyright holder sues, they’ll have to name the human bits (melody, lyrics, arrangement, performance, creative editing) and show a real access path—not just “it sounds like mine.” Defendants who can show independent creation (their own sessions/prompts/prints) will beat a lot of these cases early. SUNO users have a shield.

Chuckie Pierce's avatar

Spent a few days working with DeepSeek to make sure mine are defensible.

Output a process document that explicitly shows transformation. Saved the documents to my cloud.

Plus, YouTube's copyright filter doesn't pick up on it.

Mark Pierce's avatar

I wonder if SUNO could offer a documentation service to registered users.

Chuckie Pierce's avatar

Possibly. Depends on the use case and how much it intersects with fair use.

Piping output from Claude to Suno with logging could work in those relevant cases.

Rob's avatar

Kinda OT but I really want to hear that "Hit Me Up" banger again

Jefferson Kim's avatar

Here's timbalands Tata AI collaboration. A big tell is the lip syncing. It's still been very difficult to get it working without feeling uncanny with AI. Once someone can breakthrough that barrier, it'll be a major milestone.

If you look at most lipsync AI content, you can only show 5 seconds, no more than 10 before your mind just starts revolting.

https://youtu.be/0SFPs1EXprU?si=TG47DJ_qH9P3amav

Chuckie Pierce's avatar

Literally producing an album in about a week from an humorous concept that turned out very well.

Got another album idea from recent news. Got Claude doing the lyrics.

As bad as listening to the same song on repeat to make sure there are no errors happens to be, that has to be worse than dealing with musician vagaries.

It allows people who never learned or couldn't learn to sing or play an instrument the ability to produce music.

Black's avatar

"As bad as listening to the same song on repeat to make sure there are no errors happens to be, that has to be worse than dealing with musician vagaries."

AI is never going to show up six hours late and high as a satellite in no condition to play, go to jail for three months on a drug charge, or suddenly skip town because it got some random girl pregnant after a show last month.

Forget about salaries and royalties, the real financial saving with AI is all the secondary costs of hiring musicians.

Julie C's avatar

Another nice thing about AI singers, they enunciate.

To my ear, the line about "law and order" sounds like it's being pronounced "lore and order," which makes my brain want to play with the meaning. Sometimes a little ambiguity is a good thing.

Vox Day's avatar

This is very far from the final mix. I just threw it together to make the point. But yes, I noticed that too.