Products > ChatGPT/AI

Audacity AI plugins can now Deconstruct pre-recorded Music

(1/1)

jonovid:
has anyone tried the new Audacity AI plugins
free AI plugins can now deconstruct sound into its pre-recorded parts.  :o

Free AI Audio Tools You Won't Believe Exist
go to 7:33 on the youtube video.
 

John B:
I'll have to check them out on a Win11 virtual machine as they don't appear to support linux.

From the outset I can see this technology is the future and I don't really like it. From a professional musician and audio engineer standpoint, the noise reduction is probably the most impressive implementation of the technology, as sometimes you deal with source material that isn't recorded in a controlled setting.

The stem separation function was impressive, but the results had too many artefacts. However, the technology will obviously improve to a point where that isn't an issue, the question is what the point of the tech will be. I think it will mainly be used by "remixers" to make cut-and-paste music pieces.

Lastly, the AI audio generation is by far the most objectionable. As above, even though it sounds pretty bad at the moment, the tech will improve to the point where that isn't the issue. I don't agree with it on a philosophical level, and it basically result in the atrophy of human creativity.

I can't remember the exact composer, but one of them talked about the idea that every aspect of sound and music in a composition that can be controlled, should be. AI music generation seems like the antithesis of that. Even ignoring issues like overall sound quality and artefacts, from a compositional standpoint, most of the AI music I have heard has a "background advertisement music" or "corporate presentation" type of sound. Just like AI generated visual art has taken on a number of signature qualities, so too music seems to as well.

The depressing thing is that due to market forces and corporate influence, this type of audio visual media will dominate what most people experience in their day to day life.

SiliconWizard:
Can AI deconstruct people too?

David Aurora:
Moises has actually be doing this for years now. Used to be pretty clunky where you had to upload to the server and fuck around a bit, now it's available as an app for like $5-10 a month (or free with less features I think?). Splits into multiple tracks (you can choose how it separates things) and also works out chord sheets and lyrics. Pretty amazing really.

I've used it a lot to clean up/modify samples for some albums I worked on, but I actually really dig just loading performances I love into it too so I can zone in on parts I'm curious about

AndyC_772:
I've been using RipX DAW to do this for a few months now. It runs locally on my PC, taking about 5 mins to separate out the tracks in an average length song on my laptop, and the quality is superb.

I'm learning to play drums, and it's a fantastic learning tool for that. I can take any track I like, listen to just the drum part to analyse and practice, then play along to the entire song minus the original drummer.

That's not a bad thing for real, human generated, artistically meaningful music in general. Quite the opposite. It's an amazing learning and practice tool for a musician.

Navigation

[0] Message Index

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod