Found this on Youtube
https://www.youtube.com/watch?v=gTmq...af1B4_&index=6
Anyone know anything about it?
The singer sounds like Jon, but the instruments sound nothing like Yes
Maybe AI ?
Found this on Youtube
https://www.youtube.com/watch?v=gTmq...af1B4_&index=6
Anyone know anything about it?
The singer sounds like Jon, but the instruments sound nothing like Yes
Maybe AI ?
It's AI.
How would you actually get AI to do that? Do you just say "hey give me something that sounds like a Yes album?" ...and it spits that out?
Or do you actually have to give it some parameters: lyrics, chords, melodies, etc., plus I'm guessing a sample of Jon's voice? And then give it creative input over several iterations such that this is the final product?
I use AI to generate computer code. Sometimes it actually works, sometimes it provides ridiculous nonsense, sometimes it is just plain wrong. But it gives me a good place to start from. Often I can tell it what is wrong and it improves the code. Sometimes it gets offended and just shuts me out: "please try asking about a different topic". So I'm trying to picture if this music is created in an analogous way.
The music is surprisingly competent if completely unmemorable. The lyrics are hilarious. They sound like Yes but are completely banal.
Certainly interesting. Quite dense. Vocals actually sound pretty decent, especially the lead "Jon" voice and the way they all harmonize. I occasionally hear a Chris Squire voice in there as well. Lyrics are pretty dumb (that was expected). Guitar is somewhat Rabin-esque. Keys are rather generic. Drums sound terrible. Thanks for posting!
I wonder if Jon Anderson knows about this. It sounds a lot like him. Weird. Maybe he can steal it and make a new Anderson/Stolt album with it.
Do not suffer through the game of chance that plays....always doors to lock away your dreams (To Be Over)
It may sound rather generic, but the result is quite impressive (and scary in a way).
I’m actually too scared to listen
There’s way too much good new music to entertain me, without giving my precious time to this AI nonsense.
I review and blog at www.lazland.org
It was likely made with Udio or similar. So essentially just text prompts and endless iterations until it lands on something you like - no need (or even means?) to specify chord porgressions or anything. I've had a play around with it out of morbid curiosity and it is both terrifying and reassuringly crap at the same time - like most generative AI.
Udio ignores prompts asking for specific artists but that can be circumnavigated to an extent, though I'm not sure how the user managed to get such an uncannily good approximation of Jon's voice. Unless they have access to other tools/features not available with a free account?
Either way, whoever created this still put a baffling amount of effort into producing something utterly pointless and mediocre. I wouldn't mind if they were up front about it being AI, but I hate the thought that people might actually be fooled by it.
Yeah this is the thing. In their description they present it as 'lost tapes' type of thing just upscaled by AI when in reality the entire thing is a lame AI creation.
The music is likely generated by Udio or similar indeed, it has that same meandering 'yep sounds like -genre- but with nothing at all interesting happening' vibe, and the awful sound quality as well.
I feel like there must have been something slightly more manual going on with the vocals however as I'm not sure those AI music generators can imitate specific voices yet, that's some separate AI tech afaik.
Perhaps they generated 'Yes-like' songs in Udio and then split the vocals and instrumental apart using other AI tools, and then converted whatever voice was generated into a Jon Anderson voice using one of the voice converter AI tools.
Nothing to worry about, it's not at all entertaining lol it's actually actively frustrating to listen to how banal and aimless it is.
Nobody was talking about Davison, what's the point of this?
If this is where AI is at now, where will it be in ten years?
Agreed. There's plenty of new music out there that deserves our attention rather than Yes AI fakery. I listened to quite a bit of it, if I was Jon Anderson I would sue to get his voice removed, no idea how to go about it though and there's the problem. I very much do not approve of this AI crap.
I expect you're right, but I feel like we're a long way off hitting that wall. I find it hard to imagine a future in which humans abandon creating music for themselves in favour of AI-generated mulch, but equally I think it's naive to assume that this tech will not have a direct impact on the music industry, especially when it comes to composing for film, TV, video games, advertising, etc. (Not accusing you specifically of doing that, to be clear!). I also think it is only a matter of time before a major pop artist openly admits to using AI to create demos/generate inspiration for songs, even if they are then polished and professionally recorded.
Yeah I have a feeling the primary place where humans will actually be replaced by AI in terms of music is when it comes to synch, like movie and TV music, ad music, game music etc. like you say. A place where a lot of money is spent to license music, and unfortunately that is one of the few big money makers left for musicians. Not for long perhaps...
The thing with AI 'composition' is that it's not really composing songs. Composing isn't really a part of the process - it is not creating it as written sheet music or even MIDI then turning that into a song. It is directly creating waveforms / 'final products', 'mixed' and 'produced'.
Because of this and because of its data-set, by the fact its basically doing the equivalent of a phone's texting 'autocomplete' in terms of music, with the current model it will always create very generic and stale music.
It will always create some sort of middle-of-the-road aggregate of the music it was trained on. Safe bets. The things that all the different music it was trained on (at least relating to the prompt its given) have in common. So nothing quite unique tends to be generated.
The types of people who are creating these music AIs and are really into them are also the types who are already listening to these AI creations and thinking it's actually good music, so that tells you a lot about their barometer for creativity and musical value.
The thing about these AIs is that it makes you feel like there is something creative happening, that somewhere beneath the surface there is some sort of creativity in there. But really they are just imitations of what they have been fed, just like the visual art generators. It will inevitably always remain lesser and derivative of what it's been fed (at least with this current generation of generative AI models) until such a time as it can be taught or reconfigured to actually process and understand the musical elements of what it has been fed, break it down with some sort of actual musical understanding and then create new pieces via composition.
I think the pop industry should be more worried about AI than the smaller genres, but it's certainly powerful tools.
The below (Chinese YouTube style link) used to be on the real YouTube - but Queen had it removed as they don't have an issu with Freddie's AI voice being used to sing Elton songs or Whitney Houston tunes, but don't allow their own songs to be manipulated.
First time I heard it - I was stunned at how very very close it was to Freddie.
https://m.bilibili.com/video/BV18N411D7Cb
Sent from my NE2217 using Tapatalk
Death inspires me like a dog inspires a rabbit
I saw this a few days ago. Might be of interest. The music itself is crap but his analysis is good.
It's kinda scary actually. All that music he reviews is all AI.
Still alive and well...
https://bakullama1.bandcamp.com/
Bookmarks