Is the Future of Music a Chip in Your Brain?
wsj.com
wsj.com
As an aside - this feels like a "submarine" post (http://paulgraham.com/submarine.html). I can't tell what they are selling though.
I could easily make such an algorithm that would work well for myself. Tag songs with a mood, have different modes that switch what kind of mix I want. Add in the ability to, in the player, designate certain songs to not play in the currently playing mix, and I think I could iterate my way towards never having to skip a song again. It's on my list of things to build 'one day'.
But that's a far cry from being able to generalize over what people that aren't me would be like. It's a hobby of mine to ask random people how they consume music. We'd essentially need the power of a full-on programming language to make object frameworks with customizable predicate logic for each individual person that they would have to learn and implement, to improve upon the skip button. The domain is that hairy.
Mind reading is not for the faint of heart.
* whether machines will be capable of composing some kind of appealing music. This is already possible, and likely to get better/broader.
* whether machines will get so good at this (on demand, no less) they will crowd out contributions that are largely human-crafted. This seems uncertain to me, maybe more likely at the commodity and conceptual ends of the market, maybe less so in craft-niches.
* whether music is really a personal experience or a social one -- or, since it's obviously both, to what degree people want both things when.
The usage of machines in music currently leads to a proliferation of basically glammed out pop rock with pitch perfect voices metronome timed beats and wildly predictable song anatomy. The author mentions that Sinatra's voice has been reconstructed and has released new hits... But isn't 90% of Sinatra's appeal the broken, manly _imperfection_ of his voice?
In short, I hypothesize that machines could adequately "fill the void--" ie. play your running music, play your driving music, give you some background tunes to study or code to. But to anyone that actually pays attention to music it will be wildly subpar. Imagine a machine putting off the same lyrical heat as Kendrick Lamar or Ghostface? How can a machine possibly get on a novel level like that when it's basic coding is necessarily to average the means?
If I were rich I'd establish a big fat prize for this technology, seriously.
We can be fairly sure that when you hear music in your head, nothing is actually vibrating in your head to produce the sound. My guess is that you have a bank of "instrument memories" and "note memories," and you can imagine new music by smashing a note memory into an instrument memory. (Similarly, try to imagine President Obama saying "My fellow Americans: even electronic brain pancake crystal elderly." It's surprisingly easy to combine your memories of his voice with your bank of known words.) But what guarantee is there that my instrument memories are the same as yours, or that they are stored in the same way? The mapping is not clear.
* to give this putative "plays the music in your head" app a terrible name
http://news.berkeley.edu/2011/09/22/brain-movies/
And you have to consider that nonexistent music playing in your head has a special "unfinished" quality that might make it impossible to just translate into sound. My take is that when you process sound, it goes through multiple layers or depths of analysis, and imagined music takes place at an already deep level of analysis - could be very problematic for this project.
Haaaapy b.. what, how did that go again ?
To help you understand, there are two ways, one is using so called 'expert systems', this is computer algorithms guided by humans, and other would be artificial intelligence, where computers completely generate music on their own.
For the most part, ES is done today and works well, AI can be done for ambient music. Once we plug our feedback, how we feel and what kind of music we would like to listen to, that is when things will get really interesting.
Surgery is invasive, dangerous, your body is corrosive, putting stuff in your brain will have side-effects and what are you going to get out of it? Music to suit your mood?
I totally agree with Nick Bostrom[1] on this one, it's not happening any time soon.
[1]http://www.amazon.com/Superintelligence-Dangers-Strategies-N...
Anyhow, I think I understand where your skepticism originates, it make sense, I think a lot of smaller advancements would allow us to come closer to this.
Ableton has a small suite of algo-comp packs. And there are things like WaveDNA's Liquid Music.
I'm not aware of anyone making complete commercial tracks using ES - at least, not with a high profile career. (I suspect Bjork does a bit of this, but I'm not sure how much. Holly Herndon certainly does, but isn't very mainstream.)
Strange fact: people have been attempting algo-comp for decades, and it remains incredibly obscure and ignored, even though there have been literally hundreds of languages, toolkits, and frameworks. That's partly because most of it isn't very musical - at best it sounds like a random mash-up of music slices with no overall sense of form or mood, and at worst it sounds like musical nonsense.
It could also be because successful systems become so complicated they're only really useful to one composer. No one else has the time or the patience to learn them.
But more, musicians seem to have a block about the idea - e.g. WaveDNA have literally had a lot of hate for LiquidMusic.
Even though many genres are very formulaic, abstracting those formulas in a useable way turns out to be very hard, especially if you're aiming for a definite mood instead of a random musical squiggle. Most people who make music would rather do it by hand than have a computer do it for them.
Quick google found this, looks like what I was referring...
This is what I was referring to