Matthew Sheeran: Technological Frontiers of Music 

Your channel is so good. I've been visiting here since some 'understanding Schoenberg' kind of videos. Interesting topics!!

Very interesting topic and conversation, looking forward to the full podcast. What Matthew Sheeran mentions, about Zappa recording the bassoon at a slower tempo and then speeding it up to keep the human element, was something that zappa did very often during his early albums. An interesting example is the track "Zolar Czackl" from the Uncle Meat album where different (instrumental) material are recorded at a slower tempo/pitch and then sped up (using half-speed overdubbing on tape of course!) to match a precomposed rhythmic structure. Regarding the discussion on the technological frontiers and AI my take on that would be that our boundaries nowadays are not situated in technology (how fast something gets processed or how much data can we store etc..) but how to learn to interact with it. Most "experimental" musicians nowadays sound very similar to each other because they end up being just end-users of particular presets , effects without testing their creative possibilities. I believe that one of the creative problems of musicians/composers working with AI would possibly be for example to learn how to write the "correct prompt" for their respective creative situation - sparking the right amount of imagination to push something musically into places you cant imagine.

Although the frontiers of music technology seem extraordinary to us today, they remain under the domain of human expression. It is still humans who generate the rules that produce unexpected behaviour. But so it has always been with technology whether it is an AI or a counterpoint rule. The question for me is: will machines ever be able to desire to express themselves artistically? Will they ever be able to desire to produce music? This is the question that makes and will make the difference for me.

I think there's an interesting blind-spot here. You ask what the frontiers of music technology are. Matthew Sheeran when asking about Zappa implicitly assumes that machines can't equal the expressiveness of human players. Then you both start scratching your heads mystified where technology will go in the future. The obvious way for it to (continue to) evolve is to get as good at interpreting a score as human players are. At the moment, you are still thinking that either a human player must bring expression through performance. Or a composer must meticulously program it in with a lot of explicit control data or key switches. But I'd bet that the next thing for AI is for it to be trained to map from scores to musical performance. In other words, you ought to be able to give a bassoon score to an AI bassoon plugin, and have it produce a bassoon audio which is as expressive as a human would play it. Perhaps informed by seeing (or even hearing) the other parts.

Are composers and/or aspiring composers a kind of modern Don Quixote? That is, an individual who aspires to succeed in the contemporary world by practicing a profession from the past, similar to the way Don Quixote pursued the role of a wandering knight.

I love this channel but here I was a little bit shocked. Music is not about picking the "best" iteration of something. Music is about finding the one necessary iteration. Music is about the flow of sounds a human brain can think of and a human body can produce. AI will bring out an idiotic sense of "perfection" in people, that's one of the reasons why I'm afraid of it.