completely A.I. generated songs with singing are here (Udio/Suno)

Anything about MUSIC but doesn't fit into the forums above.
Post Reply New Topic
RELATED
PRODUCTS

Post

igorius wrote: Sun Apr 14, 2024 6:49 pm If I stop subscribing to a plugin vendor, I can no longer open my old projects, unless I resubscribe.
Render to stems prior to cancelling, and you can continue to fiddle (within limits)
We are the KVR collective. Resistance is futile. You will be assimilated. Image
My MusicCalc is served over https!!

Post

I'm not sure that's necessary, though, is it? My only experience is with Arcade and all they do is freeze the GUI so you can't edit anything, but all your old projects open and play as normal, whether you're subscribed or not.
NOVAkILL : Legion GO, AMD Z1x, 16GB RAM, Win11 | Zoom UAC-2 | MPK Mini+ | Studio One 6.6
ARP2600, ARP Odyssey, OB-EZ, SEM, OB-1, Prestige, GR-8, JP6K, Union, Saurus,
Invader 2, Olga, TRK-01, BA-1, Thorn, Spire, VG Iron 2

Post

Check this out. If it was on a Depeche Mode album, you wouldn't even notice the difference.

 https://www.youtube.com/watch?v=7U57R_icuOg
NOVAkILL : Legion GO, AMD Z1x, 16GB RAM, Win11 | Zoom UAC-2 | MPK Mini+ | Studio One 6.6
ARP2600, ARP Odyssey, OB-EZ, SEM, OB-1, Prestige, GR-8, JP6K, Union, Saurus,
Invader 2, Olga, TRK-01, BA-1, Thorn, Spire, VG Iron 2

Post

BONES wrote: Wed Apr 17, 2024 1:02 am Check this out. If it was on a Depeche Mode album, you wouldn't even notice the difference.

 https://www.youtube.com/watch?v=7U57R_icuOg
That's pretty-much, jaw dropping amazing. 8)
I'm not a musician, but I've designed sounds that others use to make music. http://soundcloud.com/obsidiananvil

Post

The vocals on that Depeche Mode A.I. track are stunning. I don't have inside information but I am pretty certain Udio ran millions of commercial songs (including Depeche Mode) through stem separation and trained their models on what was extracted. The vocal stems must have been given special attention. I have continued to tinker around with it. It has occasionally produced amazing, professional sounding vocals that sound totally human to me. The results can vary dramatically, even with the same prompt. It's sort of like a slot machine. Sometimes you hit "Create" and you get something adequate, other times, you hit the jackpot and it produces magic.

I think I know where this is heading. This is going way beyond musicians using A.I as a tool. I predict Spotify will start generating A.I. music and putting it directly into algorithmic playlists (which are based on listeners' tastes). Spotify will love this because it can reduce dependence on those pesky artists and record labels that want to get paid. Spotify can just license Udio's tech (or acquire them outright). I think the results will soon be good enough that average Spotify listeners wouldn't mind listening to A.I. generated stuff alongside human artists.

I imagine Spotify would have a special icon to indicate which tracks are A.I. generated and give you an option to opt out of A.I. music completely but most people won't bother with details like that. We are living in interesting times.

Post

Yeah, I'm surprised all the big record companies haven't jumped all over this stuff. After all, they've been manufacturing stars out of nothing for decades.
NOVAkILL : Legion GO, AMD Z1x, 16GB RAM, Win11 | Zoom UAC-2 | MPK Mini+ | Studio One 6.6
ARP2600, ARP Odyssey, OB-EZ, SEM, OB-1, Prestige, GR-8, JP6K, Union, Saurus,
Invader 2, Olga, TRK-01, BA-1, Thorn, Spire, VG Iron 2

Post

Crazy. Suno wouldn't allow entering an artist's name though. It refuses to create a track as soon as it recognizes the name of a famous artist in the prompt. Also, I might be wrong, but have the impression that Suno has a library of voices they use, that might be from singers that they've gotten the permission for AI training from. When I stick with a genre, I get results with a limited set of voices. At least that's my impression.

Post

igorius wrote: Wed Apr 17, 2024 3:57 pm Crazy. Suno wouldn't allow entering an artist's name though. It refuses to create a track as soon as it recognizes the name of a famous artist in the prompt. Also, I might be wrong, but have the impression that Suno has a library of voices they use, that might be from singers that they've gotten the permission for AI training from. When I stick with a genre, I get results with a limited set of voices. At least that's my impression.
We managed to get modern Metallica/Hetfield by saying "1991 mainstream metal"
.................................
"Hell is other People" J.P.Sartre
.................................

Post

> Check this out. If it was on a Depeche Mode album, you wouldn't even notice the difference.

that's debatable. Lyrics give it away if not the audio quality :-)

but wow, they got dave gahan's voice! haven't heard anything quite as obvious yet.

I thought this was quite a good Ian Curtis (suno) :

https://www.youtube.com/watch?v=4UY1HHoCVaw

Post

hibrasil wrote: Wed Apr 17, 2024 4:36 pm that's debatable. Lyrics give it away if not the audio quality :-)
The lyrics in the example are bad but you can type in custom lyrics into Udio or generate them using a different A.I.

Post

This guy tests A.I. music and does a depressing, long-winded rant. I agree with his conclusion that A.I. will be able to generate high-quality stock music for a fraction of the cost it takes to hire a human composer. As a result, most stock music creators will be the first to be put out of business. And eventually most musicians who earn a living from recordings will need to find different jobs.

https://youtu.be/wEQ9Vg2YKcU?si=ci-NqtUN39s-sr8i

Post

this guy is great

Post

Amazing times indeed. Humanity is just millions of lemmings gladly jumping off a cliff while staring at iPhones at this point. Bring on the Terminators I say, this goose is cooked. :lol:

Post

lotus2035 wrote: Wed Apr 17, 2024 7:12 pm Amazing times indeed. Humanity is just millions of lemmings gladly jumping off a cliff while staring at iPhones at this point. Bring on the Terminators I say, this goose is cooked. :lol:
Here you go:

https://www.youtube.com/watch?v=29ECwExc-_M

... and of course making it sing is trivial... it could probably dance too (very flexibly!) like their other robots:

https://www.youtube.com/watch?v=kpeODvGJE1Q


Edit: or for a little more flavor:

"Scientists unleash army of remote-controlled cyborg cockroaches in desert"

https://www.independent.co.uk/news/scie ... 29993.html

Post

I can see there will be problems with misuse and such, but I really look forward to using AI in a DAW-like environment where you could build up a virtual jam band.

If you could export the tracks as MIDI for further refining and arranging, replace the instruments, get a score in notation or tab, etc — then I think it could be everything I thought music and computers was going to be when I first became interested. All I ever really wanted was a cooperative jam band that knows more about music than I do in a recording environment with unlimited patience for takes and unlimited editing potential. Perhaps AI will eventually even be able to break down the theory of the generated track and suggest how various modes , scales, arpeggios, triads, chord substitutions, etc might be incorporated into the tune. In that way I think it could be a very positive thing for human music.

Post Reply

Return to “Everything Else (Music related)”