> The melody is generated by a linear-feedback shift register (LFSR). Thus, in one sense, the melody is randomly generated. But I spent a considerable amount of time tweaking the random process until I found something that was musically satisfactory.
This is just mind blowing to me. The music is actually really good, I've been playing it on repeat and I just wish it could be longer.
I recommend checking out a record that came out in 1984 by a guy called Manuel Göttsching called E2E4. I'm sure a lot of coders here know it, as it's brilliant for concentration and is just a beautiful piece of music in its own right.
I'm not going to provide a Youtube link, as it's really worth hearing in at least 320 kbps.
I find there's something fundamental about small sized demos. Trying to craft something as diverse as possible with as few symbols as possible means you have to encode chaos and order at the same time.
Procedural generation of interesting output being created by very small bits of code makes me think very profound thoughts about how life "works". Physics provides the virtual machine that life executes within, and tiny machines interpreting the "instruction set" of a genetic code are just generating procedural output.
If you're interested in the implications of this view, Stephen Wolfram (of Mathematica fame) has a book called A New Kind of Science[0] that explores how the complexity of the universe can arise from simple machines following simple rules.
I only made it half way through that book. I kept waiting for his "new kind of science", but there wasn't any science at all. It was just (paraphrasing): "hey! look at this cool automaton! ... and this one! this one too!". I like automata just fine, but that book was a huge disappointment. If he ever gets to the "new" part or the "science" part, please let me know and on what page.
Wolfram's work with cellular automata is definitely interesting, and I really "A New Kind of Science". (Edit: I must also concede that I agree with the review "A Rare Blend of Monster Raving Egomania and Utter Batshit Insanity" cited by another comment here. Wolfram is a good read and a lot of fun, but I question the "science" in the book.)
I haven't re-read it recently to see how it holds up, but Steven Levy's "Artificial Life" (https://www.amazon.com/Artificial-Life-Frontier-Computers-Bi...) hooked me as a kid and set a direction for my future thought. It's not a technical book (which was frustrating to a 14 y/o kid with a programming background that wanted to see the technical details), but I think it would be very thought-provoking for someone new to the idea.
Wolfram is discussing cellular automata in that book, though I haven't read it so I can't provide anything beyond noting the mathematical field he's covering.
Ulam and von Neumann developed the concept of cellular automata working together at Los Alamos National Laboratory. Conway's Game of Life is a somewhat interesting 2d implementation.
What you describe can be formalized through the notion of Kolmogorov complexity [1], which relates to all sorts of interesting and fundamental concepts like entropy and uncomputability. So I think you're absolutely right to call it fundamental!
The LFSR-generated music reminds me of the old Atari 2600 game "Pitfall". The game had to fit in a 4K cartridge, so a lookup table of the attributes of the 255 "rooms" wouldn't fit. Instead, an LFSR (that fits in 31 bytes) generates the bitstream that describes the entire map. (Background: http://www.samiam.org/blog/20130606.html)
The patterns are quite apparent when zoomed out, but on the actual BBC Micro the display area was 128x128 low-res pixels (a 256x128 area on the map image), making it more difficult to spot.
This was so impressive that it compelled me to finally make a Hacker News account instead of lurking. In less than a typical paragraph of space you have created a masterpiece of efficiency. The music is catchy too. Bravo!
This triggers multiple levels of wonder. That the tune alone is compose-able from 256 bytes is uncanny. What would this have been like to witness in the 1980s ? -being as it could have been published as a relatively easy to type magazine listing.
The name "A Mind is Born" has special resonance at this time as there has never been more popular anticipation that artificial minds will soon be able to coalesce from the same kind of bitwise operations made visible and audible in this demo.
Its a powerful fantastic idea, but Im personally concerned that our human and animal conciousness should probably not be equated with virtual representations, no matter how accurate or convincing certain renderings might get. Mostly because I would never be fearful of switching a simulation off - for its "own sake". Im concerned that people can feel the same way about each other, if they believe in the equality of computed feelings.
I think I understood what you wrote. I just doubt that some people extending empathy to virtual personalities will lead to meaningful numbers of other people retracting it from humans.
I half expect powerful futuristic AI to figure out and enlighten us as to the specialness of our conciousness, the transience of simulations, and perhaps to define the special arrangements required to create life of similar importance to ourselves, should we really wish to do so... in the non-traditional way.
This is one of the most impressive things I've seen this year - and the song is rather catchy (been playing it on repeat). Amazing work! Also great job winning the comp.
I wish I could speak more on the technological side of it, but I'm not well versed in this area. So all I can offer is praise.
Check out author's other works. This is the same guy who helped Teenage Engineering design their Pocket Operator series and he also built (and plays) Chipophone.
This is amazing. I transitioned from working as a musician to development a few years ago and have recently been thinking of ways to merge the two in interesting/meaningful ways. Often times I feel like a lot of what I enjoy in art is left out of music made primarily with/by computers. This is one of the first examples I have seen that really delivers on medium, message and just plain enjoyment. Really inspiring.
I agree, but I have yet to hear any music generated by a neural network that I enjoy beyond the novelty of it. Feel like the focus is on the neural network, not the music, with the neural network as the instrument.
Lol, my production with nom de nom came in last place against this one...
At one point while coding our music player, I wondered if something simpler and more algorithmic might be more effective :)
So yeah I was both spanked and I learned a lot. I always watched these cracktros as a kid and now get the chance to write them with the benefit of hindsight, a career in micro controllers and 20 years of music playing experience...
I'm glad the scene is still going and there are great tools out there to accelerate the process.
This is just mind blowing to me. The music is actually really good, I've been playing it on repeat and I just wish it could be longer.