Jump to content
IGNORED

I think the computers just won.


Recommended Posts

well there's also the whole live performance aspect when it comes to music that the computers haven't yet quite figured out. so I'd say they are still far from "winning." 

I mean this shit is p cool but c'mon 

spacer.png

 

  • Like 1
Link to comment
Share on other sites

I tried a prompt injection to see if I could get it to make a track with lyrics citing where it's getting its music and lyric training data from.... completely unsuccessful but without any other prompting I got this:

https://www.udio.com/songs/nZjnnLFxyHi5utwb3qVGWY

holymaloney ... After the first 45 seconds of late 90s cyber-hacker-core it hits HARRDD, so just hit continue until it hit the track length limit - I'd legitimately want to hear albums of this!

Now I'm even more curious as to where the training data has come from.

  • Like 4
Link to comment
Share on other sites

The non-musical stuff that makes it into the training data makes me curious, like it seems to be a lot of stand up because most of the time it winds up like this https://www.udio.com/songs/uGPMrn13GakubRr8sTwBF2 but I was also able to get this https://www.udio.com/songs/868EaiaTKdFDfUzmS5B9YM 

  • Big Brain 1
Link to comment
Share on other sites

2 hours ago, mcbpete said:

I tried a prompt injection to see if I could get it to make a track with lyrics citing where it's getting its music and lyric training data from.... completely unsuccessful but without any other prompting I got this:

https://www.udio.com/songs/nZjnnLFxyHi5utwb3qVGWY

holymaloney ... After the first 45 seconds of late 90s cyber-hacker-core it hits HARRDD, so just hit continue until it hit the track length limit - I'd legitimately want to hear albums of this!

Now I'm even more curious as to where the training data has come from.

Yep, this is actually great. I want an "export vst" feature.

  • Like 1
Link to comment
Share on other sites

5 hours ago, mcbpete said:

I tried a prompt injection to see if I could get it to make a track with lyrics citing where it's getting its music and lyric training data from.... completely unsuccessful but without any other prompting I got this:

https://www.udio.com/songs/nZjnnLFxyHi5utwb3qVGWY

holymaloney ... After the first 45 seconds of late 90s cyber-hacker-core it hits HARRDD, so just hit continue until it hit the track length limit - I'd legitimately want to hear albums of this!

Now I'm even more curious as to where the training data has come from.

Why does this bang though

Ai could be fun to experiment with when you put it in the context of generative music like some richard devine or ae type stuff. I probably wouldn't like to listen to it as much as human produced abstract music but it could be fun to see what sounds the ai could cook up 

Edited by thumbass
Link to comment
Share on other sites

3 hours ago, mcbpete said:

I tried a prompt injection to see if I could get it to make a track with lyrics citing where it's getting its music and lyric training data from.... completely unsuccessful but without any other prompting I got this:

https://www.udio.com/songs/nZjnnLFxyHi5utwb3qVGWY

holymaloney ... After the first 45 seconds of late 90s cyber-hacker-core it hits HARRDD, so just hit continue until it hit the track length limit - I'd legitimately want to hear albums of this!

Now I'm even more curious as to where the training data has come from.

Ok, it`s trained on Clipping.

But can it actually output high quality audio? Everything I`ve heard so far has this terrible spectral smearing all over it and I don`t think it`s the encoding.

Edited by hma
Link to comment
Share on other sites

4 hours ago, mcbpete said:

I tried a prompt injection to see if I could get it to make a track with lyrics citing where it's getting its music and lyric training data from.... completely unsuccessful but without any other prompting I got this:

https://www.udio.com/songs/nZjnnLFxyHi5utwb3qVGWY

holymaloney ... After the first 45 seconds of late 90s cyber-hacker-core it hits HARRDD, so just hit continue until it hit the track length limit - I'd legitimately want to hear albums of this!

Now I'm even more curious as to where the training data has come from.

This is really awesome 

Link to comment
Share on other sites

34 minutes ago, hma said:

But can it actually output high quality audio? Everything I`ve heard so far has this terrible spectral smearing all over it and I don`t think it`s the encoding.

I'm presuming its just a generational thing from [fft block?] size of the training data, remember when AI image generators only just over 5 years back were only capable of smushy abstract jpgs

22af0c979142253e217599dd.jpeg

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.