Music scientist Phillip Dorrell [0] has argued for the existence of currently hypothetical "strong music," a class of musical stimuli presumably discoverable by strong AI.
The idea makes sense if you accept the concept of "intelligence explosion." Any property, in this case the rewarding effect of acoustic stimuli in humans, can be powerfully maximized. There must exist patterns in music-space that would have profoundly greater impact on human minds than those our low-wattage brains can find. So through a really powerful search and optimization process that can more efficiently explore remote, undiscovered regions of music-space, we get a "music explosion."
What these songs would sound like is the real mystery. Would they sound anything like the music we're familiar with? Would they lead to musical wireheading?
It also seems a bad idea to measure musical goodness by, say, how many times humans will replay a certain audio file. If you use this measure, I don't think you'll end with what you want at all.
This assumes music's value and power are absolute and independent of the listener's experience. But most evidence points to the opposite.
For one, music is culturally dependent and music from some regions of the world sound bad to people from other regions. Secondly, music becomes more enjoyable the more you listen to it. Yes, people certainly replay the songs they like, but replaying them makes them like them more; it's a positive loop that that feeds itself. And also, emotions derived from music use many other pieces of non-musical information: connection with the artist / ideas expressed in the piece / personal memories / etc.
All of this makes it quite unlikely there is some absolute sound yet to be discovered in the musical universe that would wow anyone on the first hearing.
If one is closed minded to learning math or the importance thereof, mathematics will be difficult. Similarly, if you see no value in exploring the world's cultural output, those explorations should prove quite difficult.
I do agree with your conclusion however, as it is unlikely that some "perfect sound" exists. If it did, we would not have made so much music.
We have had “hit factories” for quite some time now, suggesting at least there is some kind of musical output that helps the process of wow-ness upon first listen.
Sure, that formula does not always work on the moving targets of changing times and tastes, but i would argue that we actually do have a somewhat defined framework of that absolute sound (which has been around for some decades now). It’s calles pop music, now a genre unto itself. Yeah, you may not WOW every living human on earth every time with great accuracy, but we have McDonalized music enough to small-wow millions at a time on a first listen basis, just by virtue of being made using quite narrow musical frameworks using rythm, scales, chord progressions, arrangement etc
```My name is Philip Dorrell. I currently work as a software development contractor. Although my only formal qualification is a B.Sc (mathematics) from the University of Waikato, New Zealand, I have always had an interest in the fields of science, mathematics, and various other fields not commonly thought of as being scientific or mathematical, but I prefer to approach them that way anyway.
I have a personal website at http://thinkinghard.com/, which contains articles on various subjects and also some items of software I have written.```
I'm not sure I would call the author a music scientist given a lack of formal grounding or experience in music theory, composition, and performance. This kind of conjecture is completely devoid of the context of the history of music which in many ways is the history of language and art. What makes art pleasing to consume is highly contextual and not really something that can be reduced to context free grammar.
Humans have been experimenting with the production of rhythmic sounds for aesthetic effect for millennia. Considering the extensive search through possibility space that this represents, I'm skeptical that a machine aided search could discover new local optima with a significantly greater aesthetic effect.
That's because the space of Go moves is a mechanical endeavour in itself, not very different from doing accounting, with no emotional or cultural element
This is incredible! I've wanted to build something like this for a while.
I sing in a barbershop quartet with 4 engineers, and we are constantly trying to tweak our tunings to optimise consonance of the chords we're singing. It can lead to interesting occurrences where you have to tune 2 consecutive notes differently
despite the fact they are on paper the same note.
This is interesting but I wish it included some information about the definition of "dissonance" being used. Of course xenharmonic stuff sometimes sounds out of tune just because it's unfamiliar, but also there's a temptation to go too far the other direction and assume that something must be exploring profound new harmonic space just because it sounds bad.
The idea makes sense if you accept the concept of "intelligence explosion." Any property, in this case the rewarding effect of acoustic stimuli in humans, can be powerfully maximized. There must exist patterns in music-space that would have profoundly greater impact on human minds than those our low-wattage brains can find. So through a really powerful search and optimization process that can more efficiently explore remote, undiscovered regions of music-space, we get a "music explosion."
What these songs would sound like is the real mystery. Would they sound anything like the music we're familiar with? Would they lead to musical wireheading?
It also seems a bad idea to measure musical goodness by, say, how many times humans will replay a certain audio file. If you use this measure, I don't think you'll end with what you want at all.
See also "supernormal stimuli": https://www.sparringmind.com/supernormal-stimuli/
[0] https://whatismusic.info/