Well, Miller created Max, which back in those days was more about composition than synthesis and relied on external hardware to create sound. Another fella took Max and developed the commercial product and developed it some more. During this time home computers became good enough for realtime synthesis so Miller developed PD which combined aspects of Max with realtime audio synthesis capabilities. Max thought this was pretty great so integrated pd into Max which gave us Max/MSP, Max is Max Mathews who developed the Music(n) family of programing languages and MSP is Miller S. Puckette.
I would not say that pd or Max/MSP is more approachable, they are pretty even these days unless you stick to vanilla pd with no externals which almost no one does, the main differences is that Max/MSP gives you all you need out of the box but pd can run just about anywhere including in plugins if you are that sort. Max/MSP is considerably more modern in aesthetic but PlugData[0] and PurrData[1] offer more modern interfaces for pd with a few new tricks, PlugData is a plugin version of pd (also standalone), Purrdata is a JavaScript rewrite of the interface with prepackaged externals and solid documentation. And we have a handful of other varieties of pd each which have their own slant and can be found on puredata.info.
Edit, ok not having a dozen different varieties and hundreds of externals to choose from does make Max/MSP more approachable. I was mostly speaking from the standpoint of use but I realized all I did was say that pd is an example of a major flaw of OSS in the days of analysis paralysis.
Shameless plug: for something more approachable, I've created NoiseCraft, which runs in a web browser, has fewer primitives, and is designed to be easier for beginners to grasp: https://noisecraft.app/1136 (just hit "Play" in the top-right corner).
Max has graphics support and is much used by artists and people making arty minimalist (interactive) media.
TD has more of a game-ish aesthetic. Max/Jitter is better for chopping up and distorting video streams and grunge-ifying them, with some simple renders and FX on top.
As a music system - it's in a weird place, because it's incredibly awkward if you're used to even the most rudimentary coding. A construct that takes a minute or two in Python can take five or ten times longer in Max.
But it has some decent DSP features, including custom DSP coding, so it's a good way to get certain things done. Or at least prototyped.
PureData and Max are primarily about synthesis and composition not multimedia, they are just capable of doing video as well. Both are still relevant. Different markets.
First three videos are ripe with technical difficulties but after that things get sorted out. Quite interesting so far, videos alternate between explaining the pd code base and the development of realtime computer music composition/synthesis.
I'm not sure who ever said "if you want to make music on a computer, learn DSP programming".
Making music has many different approaches, and potentially lots of different stages that can be permuted, combined, discarded, repeated. Composition, performance, timbral adjustments, editing, mixing ...
That said, tools like PureData and Max/MSP are not there just to do DSP. Consider this lovely example of generative music that has almost no DSP going on whatsoever - the entire patch is focused on composition:
That said, if you wanted to make music that sounds like Fleetwood Mac or Taylor Swift or Duke Ellington, then computer tools are probably not where you should start (though may be where you end up).
That's a really interesting piece you linked, thanks, and very impressive. I'm almost embarrassed to admit I could have mistaken (a couple of minutes of) it for some Sun Bear period Keith Jarrett! I couldn't make out how the piano sound itself is produced. Samples, I suppose, rather than some Pianoteq type modeling?
I don't know what is being used for sound generation. I don't know Max all that well, but blowing the video up to full screen, it looked to me as if it only generates MIDI. If that's correct, then I suspect either one of the modern sample-based pianos or pianoteq.
I would think of it more like: if you want to learn to play the piano, you'll need to sit down, learn some music theory and practice scales.
This advice isn't necessarily the only way to go especially if you're happy just tapping out Chopsticks, but as a general rule ... it's a proven path.
Same applies for electronic music: you need some understanding of theory, terminology and available tools, else you just end up twiddling knobs and creating "cool" effects.
It puts the user in the right ___domain for computer music, which is music/sound made through direct coding and control of software data structures.
The point of computer music is to invent new things that can't be done elsewhere with software, not to copy existing things that have already been done.
If you just want to mess with loops and plug-ins, use Ableton/Logic/Cubase/etc. They use the studio/sampler metaphor, which is completely different.
Could you clarify more? I think I agree when it comes to pd but not the other audio DSLs, pd has the unfortunate image of being a modular synth program which is not at all accurate and I have noticed many pick it up because of that image and it kills their momentum, they spend a great deal of time learning a tool that is far more than they want or need. I have not noticed much of this in the other audio DSL communities I have been a part of over the years, people have a much clearer idea of what they are getting into. But I am not sure I understand what you are saying.
First, Pd is not only about DSP. Second, it depends very much on the kind of music you wanna do, right? We all agree that one wouldn't use Pd to produce the next Taylor Swift hit.
People have done some fantastic stuff in Max but it's definitely the wrong level of abstraction for me. Too low level for dealing with music on a more conceptual level and too awkward for lower level coding compared to text based languages.
I'm expecting generative AI to blow these kinds of tools out of the water very soon.
I'm curious at what you would think of https://ossia.io (i'm developing it). It's explicitly "one layer above" Max/Pd in terms of abstraction level (e.g. the objects are closer to Ableton/Bitwig devices or TD objects than Pd/Max objects) but allows to delve down to code in various languages whenever it makes sense (JS, GLSL, math expressions, C++ even)
Another very good resource for learning DSP / computer music is the Bela course. You don't need to have a Bela for the course!
https://youtube.com/playlist?list=PLCrgFeG6pwQmdbB6l3ehC8oBB...
I am also developing a computer music language here at: https://glicol.org
Book recommandation:
Pirkle, Will C. "Designing Audio Effect Plugins in C++: For AAX, AU, and VST3 With DSP Theory." (2019).