A Catalogue of Chords Example(jpreston.xyz)
jpreston.xyz
A Catalogue of Chords Example
https://jpreston.xyz/a-catalogue-of-chords-example.html
10 comments
I don't think we'll get there with AI unless we get to point of creating general intelligence.
The AI compositional developments are pretty wild, but I think without the varied high-entropy input that comes from lived experience I'm not sure they'll ever really get there.
I've repeated this before—many (maybe most) great artist will tell you that the songs didn't really come from themselves—they just sort of felt it, they were handed it, or they reached out and pulled it down.
Some of my favourite songwriters are quite open about their process including sitting on the couch noodling mindlessly and playing with melody like it was putty (rather than an equation).
Ultimately I think it follows the old adage that if you analyze something too much it will lose another aspect of itself that isn't analyzable. You look to solve the problem of creating great music by reducing it to a problem that we have mathematical models for, then you will lose all of the input from areas we where don't yet have mathematical models.
We "don't yet know what we don't know" about music.
Definitely a fun area of discussion. I get bored fast when people try and reduce composing to such restricted values, though—except as a compositional technique unto itself. Music is math, alright, in as much as everything is math. But our mathematical model isn't advanced enough to coherently describe the experience in so few terms as people in attempted discussions about objective merits of/in music tend to reduce it.
The AI compositional developments are pretty wild, but I think without the varied high-entropy input that comes from lived experience I'm not sure they'll ever really get there.
I've repeated this before—many (maybe most) great artist will tell you that the songs didn't really come from themselves—they just sort of felt it, they were handed it, or they reached out and pulled it down.
Some of my favourite songwriters are quite open about their process including sitting on the couch noodling mindlessly and playing with melody like it was putty (rather than an equation).
Ultimately I think it follows the old adage that if you analyze something too much it will lose another aspect of itself that isn't analyzable. You look to solve the problem of creating great music by reducing it to a problem that we have mathematical models for, then you will lose all of the input from areas we where don't yet have mathematical models.
We "don't yet know what we don't know" about music.
Definitely a fun area of discussion. I get bored fast when people try and reduce composing to such restricted values, though—except as a compositional technique unto itself. Music is math, alright, in as much as everything is math. But our mathematical model isn't advanced enough to coherently describe the experience in so few terms as people in attempted discussions about objective merits of/in music tend to reduce it.
Speaking of chords: If anyone is interested I'm building a web app to browse chordsheets and create setlist (supporting offline devices).
The song format is a mixture of the ChordPro format and Markdown.
https://github.com/chorddown/chordr
The song format is a mixture of the ChordPro format and Markdown.
https://github.com/chorddown/chordr
To my (untrained) ear there is a lot of gratuitous sounding dissonance in these progressions. Not sure if an overlay of melody would somehow motivate and tie things together? In any case this is quite inspiring to learn more about harmony and creating stuff with code...
What's missing is voice leading. All of these could make sense, but if hearing it for the time, typically need the horizontal motive explicated. On subsequent listens they sound fine. Also these are hardly dissonant in today's popular music idiom.
that makes sense. I suspect my perception is also colored by my modal / non-Western music exposure: harmonies here are simpler by necessity and an "interesting chord" is invariably a major event(no pun) that needs its space and place in the sequence
In my 'non-western music exposure' I feel that we (chinese music) are not exploring the progressions well enough. A lot of emphasis is placed on melodic leading voices but largely ignores the underlying chord progression. Anything that's not overly-heard / played are deemed dissonant and "too foreign". The results are often too sweet sounding and repeating of others. I guess that's why there are not many jazz musicians from China (lol).
I only like four chords: https://www.youtube.com/watch?v=oOlDewpCfZQ
To me, in both your examples, the first three chords sound like the start of something meaningful, but the fourth one loses course.
“Learn the changes, then forget them.” -Charles Parker Jr.
The latter - having some kind of lookup chart which decides what to do next - just feels sort of clinical, and misses the feeling tone of great music. Similar sorts of complaints have been said about AI generated music and so far based on what I’ve heard, I agree.
One of the key components to writing great music is knowing when not to follow the rules. Any Radiohead song will teach you this, but they (and any good musician) know how to do it without dissonance, pushing the boundaries where they know it will work, even if it’s surprising and not “the norm”.
I guess AI will get there too. Alpha Go’s “37th move” confounded people in that it showed what felt like actual creativity, so I’d imagine there will be more and more of this in time. It’s hard as a musician or artist to concede this though, and an interesting area of discussion.