MuseNet • Hacker News Board

Re: MuseNet

#151

Post by sidthekid » Fri, Apr 26, 2019, 2:47 AM UTC

harigov wrote: post #79

BowBun wrote: post #71

Earlier quoted context omitted.

Fear not, my friend, because the execution is only a part of why music and art as a whole carries meaning for us. Consider two tracks that are identical (forget copywrite for a minute). Between one that an AI generated and a human composed, I would personally grant the human-generated version more credit and enjoy it more. The story of how art is created and the stories of the artist are as substantial to appreciatin…

What makes you think that a computer that can generate music cannot also generate a background for the "creator" of that music? It can give you all the stories that touch our hearts even more so than we can imagine.

Why generate a fake story? Why not communicate the real and moving journey of how a single note in the training data travelled through hundreds of neurons and thousands of matrices, and eventually made it past the final activation function to become a feature in the output tensor.

Re: MuseNet

#152

Post by unrealhoang » Fri, Apr 26, 2019, 2:56 AM UTC

sidthekid wrote: post #151

harigov wrote: post #79

Earlier quoted context omitted.

What makes you think that a computer that can generate music cannot also generate a background for the "creator" of that music? It can give you all the stories that touch our hearts even more so than we can imagine.

Why generate a fake story? Why not communicate the real and moving journey of how a single note in the training data travelled through hundreds of neurons and thousands of matrices, and eventually made it past the final activation function to become a feature in the output tensor.

That must be a touching story, I’m sure.

Re: MuseNet

#153

Post by throwaway66666 » Fri, Apr 26, 2019, 3:28 AM UTC

mzs wrote: post #139

> RELENTLESS DOPPELGANGER \m/ \m/ \m/ \m/ \m/ \m/ \m/ \m/ \m/ \m/ \m/ \m/ \m/ \m/ https://youtu.be/CNNmBtNcccE > Neural network generating technical death metal, via livestream 24/7 to infinity. Trained on Archspire with modified SampleRNN. Read more about our research into eliminating humans from metal: https://arxiv.org/abs/1811.06633 > More albums https://dadabots.bandcamp.com/

Despite this being a cool feat. It sounds so close to the source material (down to parts of the songs being a lofi 1-1 copy paste), that I don't see a real point of it for even existing. (except "omg death metal neural nets lol" factor)

Re: MuseNet

#154

Post by peter_d_sherman » Fri, Apr 26, 2019, 4:38 AM UTC

This thing sounds awesome! After hearing it, I am truly amazed at the level to which AI has evolved! It's good music... Heck, it's not just good music, it's potentially borderline (with some tweaks by human professionals) great music!

But, despite this potential greatness, if there's a problem, it's that this AI only produces music...

What this AI composer really needs is an AI lyricist to write lyrics for the songs it composes!

Sort of like an AI Lerner to it's AI Loewe...

An AI Hammerstein to it's AI Rodgers...

An AI Gilbert to it's AI Sullivan...

An AI Tim Rice to it's AI Andrew Lloyd Weber...

An AI Robert Plant to it's AI Jimmy Page...

An AI Keith Richards to it's AI Mick Jagger...

An AI Paul McCartney to it's AI John Lennon...

An AI Bernie Taupin to it's AI Bernie Taupin...

An AI James Hetfield to it's AI Lars Ulrich...

An AI Wierd Al Yankovic... to it's... AI Wierd Al Yankovic...

You know, an AI Assistant for this... AI Assistant...

Well, an AI Assistant to write lyrics that is... An AI "Lyrcistant"...

Come on, I know there's someone in the AI world who can do this! But it might be a bit challenging... the AI would not only have to write poetry, but it would have to match that poetry to all of the various characteristics of the music...

Not an easy task, to say the least!

But, for the right AI researcher... an interesting, challenging, worthy one!

(I think I hear 2001's "Daisy Bell" playing in the background...)

By the way... disclaimer: I am an AI. That is, An AI wrote this message on HN.

No, I'm kidding about that! But... how would you know? (insert ominous sounding music here)

Re: MuseNet

#155

Post by ramblerman » Fri, Apr 26, 2019, 5:00 AM UTC

rco8786 wrote: post #12

This seems incredibly applicable to musical scores in movies. I can imagine a product where the editor/director/someone inputs a handful of variables (mood, genre, instruments, etc) and timing requests (crescendo beginning at 30s and ending at 75s, calm period from 90s - 120s, etc) and out comes a musical score for the movie that matches up with their scene editing.

Will it ever create something that really sticks with you though. Something you can whistle or hum after watching?

Like the game of thrones melody, or star wars.

Re: MuseNet

#156

Post by quixoticaxolotl » Fri, Apr 26, 2019, 6:02 AM UTC

Permit wrote: post #66

TheOtherHobbes wrote: post #62

Earlier quoted context omitted.

The results are meaningless. This isn't chess or go, where you either win or you don't. Chopin composed for a reason, and it wasn't just an excuse to throw a lot of notes at the page. It might be possible for AI to work at that level someday, but it's not just a technical problem, and you won't be able to solve it by throwing a corpus of compositions at it. Aside from that, this still sounds like aimless noodling. It…

>This isn't chess or go, where you either win or you don't. Chopin composed for a reason, and it wasn't just an excuse to throw a lot of notes at the page. I disagree based on your following quote: > Aside from that, this still sounds like aimless noodling. It's far more polished noodling with some awareness of genre cliches, but it's still essentially aimless - and so meaningless. This sort of problem can easily be…

I think the core difference of opinion here seems to stem from a different interpretation of "meaningless".

You're arguing an academic definition of meaningful: can you design an AI that solves a hard problem?

gambler (and I think TheOtherHobbes, who you are replying to) are arguing a practical definition of meaningful: does it solve a problem people actually have?

It's neat that you can make artificial music, but actually generating music, per gambler's original comment, isn't a problem people have. It also doesn't actually add too much to culture. Essentially, the results are "meaningless" in that, even if it was successful at sounding good, what value would it actually have besides novelty?

Re: MuseNet

#157

Post by hoseja » Fri, Apr 26, 2019, 6:31 AM UTC

DannyB2 wrote: post #25

Am I remembering correctly that in Orwell's 1984 (the book), music and poetry was created by machines?

Yes, all this effort to generate music, with disappointing, lifeless results, is somewhat disturbing.

Re: MuseNet

#158

Post by davesque » Fri, Apr 26, 2019, 6:41 AM UTC

ML models that generate music continue to improve. However, to me, the result still just seems like a novelty and nothing more. Perhaps it will take much more iteration and improvement in compute power before I hear a piece of music that I find truly inspiring that was generated by an AI.

On the other hand, I've been much more impacted by visual "art" already being generated by AIs. Perhaps the musical medium is ironically harder to crack since the format is simpler, rather like how it took longer for an AI to defeat an expert goban player than a chess player.

Re: MuseNet

#159

Post by jmmcd » Fri, Apr 26, 2019, 6:44 AM UTC

CuriouslyC wrote: post #112

leesec wrote: post #104

Earlier quoted context omitted.

>Music is really complex and leans heavily on both emotions and creativity, which aren't even on the radar for AI. I definitely think creativity is on the radar for AI, see: AlphaGo. Everything we think is based on emotions is ultimately learnable.

I don't think alpha go is a good example here. It play with strange brilliance, but I wouldn't call that creativity. If you care to elaborate on why you said this I'm curious to hear your rational. To me, creativity is really about generation of "aesthetic novelty" which is hard to get from a ML algorithm that is trying to approximate patterns in training data. Eventually, there will be models trained on a wide varie…

Theorists of creativity eg Margaret Biden don't limit it to aesthetic domains. They include creativity in Go, in mathematics, even in coding. Marcus Du Sautoy has a recent intro book.

Re: MuseNet

#160

Post by TaupeRanger » Fri, Apr 26, 2019, 11:38 AM UTC

leesec wrote: post #123

TaupeRanger wrote: post #118

Earlier quoted context omitted.

Go is a game with a strict ruleset. The comparison is not valid.

One could say music is also a game with a strict ruleset. Indeed having played the piano for 20+ years now, I often think of how no matter how good I am, I won't invent a new form of music or even style of music, I am limited to the 'rules' of my piano.

One could not say that. The question is not about creating new styles or "forms" (??) of music. It is about creating compositions. I was only objecting to your use of AlphaGo as being based on "emotions" and being "creative". It is not generating explanatory knowledge, which is the kind of creativity we actually care about when we talk about AI. With current methods, you can only solve problems that have narrow optimization goals which are easily defined.