This is a pedagogical masterpiece! I knew about like 2 algorithmic approaches before, but now I know about a bunch, but more importantly I am excited to use them in my music! Thank you for your devotion to your art, it’s really inspiring!
What do you think of my vowel "solfege" diagram (more like oh-aeh), with only smooth transitions? imanuelhab.mooo.com/scale-major-vowels-imanuel.png So trippy that I had made this about a week before this video was uploaded! Man, what a perfect timing/coincidental overlapping in time and... pitch!
Many musical conventions that have been used for centuries are algorithms in some sense. For instance, the sonata form is a recipe that says something like "Write a theme A and a theme B; Start with the exposition of A, then the exposition of B, then a section of increasing tension that plays with both A and B, then a re-exposition of A, etc.". Doesn't that sound like an algorithm?
Same could be said about many genres in general. It's pretty standard in Blues music to play 12 bar blues along side a pentatonic scale melody, which limits your chord variation right from the beginning. Obviously you can break and bend these rules but they are stylistic components that define a genre. Effectively an algorithm. From there, the more you define the rules, the easier it is for a computer to create it. I suppose thats why the line between man and machine music is very grey.
This is really inspirational. It both highlights the importance of some sort of algorithmic approach to composition, but also recognises that a more formulaic approach to writing music doesn't necessarily mean that either the music will be banal, or, at the other end of the spectrum, esoteric and impenetrable. You show that with your lush string piece! Also, the editing on this is amazing. It must have taken forever!
Very insightful! One of my favourite electronic music composers has recently said that even if they could, they wouldn't share the software they created themselves to help them make music, because even though algorithms definitely play a role, they still need the composer to play them. The software, the algorithms don't do anything without the player. They're all just instruments! It is up to us to make use of them in meaningful ways. And that is ultimately deeply tied to being human.
@@DBruce this video was pretty much it! I'd love if you managed to keep it up, so nice seeing our music theory youtubers stepping up their animation game!
Thank you for making this video. I'm nobody important, but I do have a few easy tips which I hope you will consider are just one more reason for people to come here and give you a higher view count: 1) Modern composers now have so many technical resources available that what stands to distinguish them most (as not redundant with other composers) are things like their aesthetic goals and their capacity for meaningful judgment of their options; things that algorithms, alone, still don't provide all that well to listeners. Focusing on clever algorithms will tend to make a composer sound like other composers who focus on clever algorithms. But starting with way too much algorithmic data and gradually paring it down to preferred components in a preferred order will allow the composer's personality and sense of musical value to come through. Is a modern composer cheating by not using pure algorithms? Shhh... I won't tell if you don't. 2) A trick I have used to get around duplicating other people's algorithms is to take readings from a track pad and combine them with pseudorandom numbers under modulo reduction. Just wiggling your finger around on a track pad can really generate a lot of data that's not analytically reducible other than by imposing what are probably false mathermatical patterns. I'm aware of at least one composer who has used hand-sweeps of a radio telescope across the sky to pick up data. It's unlikely that what happened out there thousands of years ago relates in any discernable way to arbitrary hand motions, so the results should be pretty original in terms of being reliably random. Again, if you're an insightful editor, at least some listeners might appreciate your editorial insight as expressed in musical composition. 3) I don't think Guido's use of vowels was quite optimum. For a language like Latin, Spanish, or Esperanto (and more recently Hindi and Marathi), I prefer the idea of using 2 different (though perhaps related) arrays of 5 pitches; one for stressed syllables and one for unstressed syallables. My preferred order of these vowels in ascending pitch is U,O,A,E,I. Look at the mechanics of vocalization and maybe you'll see why. The reverse order has some merit, too, if your goal is more narrative than musical. Similarly, English as phonetically transcribed, has about an octave's worth of vowels. I default shwa to the middle of the range and spread out the others by basically drawing the shortest possible line through them on a vowel space diagram. This should work about the same with languages other than Swedish, at least. 4) The cheapest cheat code to greater melodic interest is simply to increase the proportional frequency of figures that look like "escape tones". For a complete explanation of why this works, please read (if you dare) The Analysis and Cognition of Basic Melodic Structures by Eugene O Narmour, and also this thing. www.jstor.org/stable/40285540?seq=1 The short way to combine these things is to bias melodies toward gradual escalation of nonclosure. Slowly escalating nonclosure basically makes people keep paying passive attention because it's an intensity parameter. As woodwind exercises to help condition me to improvise better, I have constructed some signatureless diatonic contours that, in a brutally algorithmic way, elaborate an "escape tone" figure with successive elaborations of similar figures; they look vaguely like fractals that expand to the right. I have no proof that this is helping me, but my subjective impression is that it is helping.
Speaking of algorithms, I've never found the purely random ones like using the digits of PI to be all that interesting as a basis to create music. I definitely dig using highly structured sequences as compositional tools. Norgard's "infinity series" reminds me of another highly ordered set of sequences: superpermutations.
Garry Kasparov's "Deep Thinking" is a great book about AI vs Human relations. The penultimate chapter is called "human plus machine" that discusses how the intelligence, and perfection of computers PLUS the creativity of humans is the greatest actor we can get. In chess, this manifests in humans making strategies, battle plans, taking enemy weaknesses into account, etc, with the machine giving them perfect predictions on the chances of each strategy succeeding, so the intelligence and creativity can be combined. I imagine that the way that composers might use this is with some kind of lick generator, so for example, if I'm starting on C and need to get to Bb an octave below, the machine could calculate the most satisfying to the ear possible way to get down there in let's say a bar, and from there, the composer could choose weather C-Bb was the best choice at all, or weather this 'perfect' lick could be changed to express something specific. This could also be done with cords, so say if I have two very, non-matching cords the AI could generate a path to that make them fit, so the composer can focus on more important things. This is pretty similar to what many of us do already, I, for instance,have a folder of drum fills or solos so when I have a vague idea of what I want I can look through what I've already made. I think the best place for AI is in this 'human plus machine' dichotomy the same way that no chess machine will ever be able to think about metagame or tilting the opponent, AI will never be able to capture experience, or put into notes nuanced emotions, it can, however, be a great resource for 'checking' yourself, meaning all the calculations a chess player is already making, such as 'are there any holes in my defence' can be answered with 100% accuracy and 0% bias, or in other words, doing the automatic calculations humans already do but without objective error. And composing already has similar automatic calculations, I don't have to think about how to resolve, I automatically default to V - I. Or in other words: AI is better at the science of music theory, but humans are better at the art of music. and together they'll probably be pretty cool
This already used for a lot of music. Not the one you hear in an opera house, but the one you hear in TV, movies, games and all around you. Programs like Band in a Box can automatically arrange the melody and harmony input as a Richard Wagner symphony, mariachi band song or Scott Joplin tune. And I'm not saying composers really shoehorn it in like this, they often completely rewrite the output , but they use these tools as part of their creative process, since it automates the boring parts of composing, letting you focus on the more interesting, creative parts...
I'm only halfway through the video, but I had to jump down to write how much I love the editing in this, and your interaction with it! It sprouts with creativity and charm, which shows that you're not really limited by your skills - or rather you perfectly worked within those limits. It's brilliant.
Speaking of composing and inspiration: 1:36 I used to go for walks in the forest with my electric guitar and that exact same little marshall amp on the shelf, just to our left of Davids ear.
Thanks, great topic! I think there are two really-good uses of algorithmic composition: 1. To flesh out details that aren’t super-important. Simplistic example: If you’re writing a 4-part chorale, you write an inspiring melody, and enticing bass line, then let an AI that understands the mundane “rules” of Common-Practice harmony fill in the details, avoiding parallel octaves and such. 2. To build an arsenal of seeds for inspiration, to break logjams in future composition: For example, have an AI write a bunch of melodies, then spend an afternoon auditing them: “crap, nonsense, garbage, ooo wait! that one has some potential; save it away! hooey, rubbish, crap…”
In the late 80s Steve Coleman who was guest artist at a school I was loosely connected to told me that he'd created a program to improvise jazz based on his own musical predilections. He said it worked well enough for famous bassist composer David Holland to tell him to get rid of it.
It's worth mentioning that counterpoint has many rules which, while still allowing freedom of the composer, are quite amenable to algorithms. There was a paper where the authors wrote a program to apply the rules of first species counterpoint to show the composer which notes were allowed according to the rules given the notes already chosen.
Fascinating exploration of music composition and algorithms! 🎵 Algorithms offer composers new tools for unique and innovative pieces, generating ideas and shaping composition structures. 🎹 It's crucial to note that algorithms enhance, not replace, creative input. 🚀 As tech advances, the composer-algorithm relationship evolves, opening exciting avenues for musical expression. 🌟
Dear Bruce. This is really an excellent video! Great. I think AI or algorithms can be used to get you to other directions you didn't even thought of. Like you said at the end of the video. I saw once a documentary in German TV about a painter who used AI like a assistant, who gives him new ideas. Looking forward to the next video. Very good and inspiring channel you have. Stay healthy, Michael
I love your thought provoking videos very much, David Bruce. Thank you. I would simply like to add that Guido d' Arezzo original hymn on which he based a new note-naming system was the "Ut Queant Laxis (Hymn to St. John the Baptist)": Ut (queant laxīs) re (sonāre fibrīs) Mī (ra gestōrum) fa (mulī tuōrum), Sol (ve pollūtī) la (biī reātum), S (āncte) I (ōhannēs) That is: Si. Later renamed by Cecile Gertken, OSB (1902-2001) as follows: Do (let our voices) re (sonate most purely), mi (racles telling), fa (r greater than many); so l (et our tongues be) la (vish in your praises), S (aint0 J (ohn the Baptist). Afterwards becoming Ti - in English speaking countries. As quoted in Wikipedia. org under "Ut queant laxis"
Really fascinating and quite unknown are two tables by William Bathe (from his "Briefe introduction to the Skill of Song") that can be used to algorithmically write or even improvise canons and canons over a cantus firmus. Peter Schubert has been using this idea for teaching improvised counterpoint, although I don't believe he has quoted Bathe.
In my 25+ year programming career, I've lived through (and worked in) several AI fads. They have come and they have gone. Composers don't have to worry, except that you may be writing less soulless corporate music. The fact is that AI still works best when it is used as an assistant or adjunct to human judgment, and for the foreseeable future, cannot be trusted as replacement for it.
At least until people start treating the work of Manfred Clynes like it's actually important. Once your AI can extract senteme contours from large training sets of commercially successful music, it's probably just a matter of time before there's music that has to be legally controlled because it makes people kill themselves or have sexually compulsive episodes.
@@joshuabroyles7565 Interesting, tell us more - what is the work of Manfred Clynes and is there music possible that could be so condusive to an emotional reaction that it must be controlled? I certainly don't believe so
@@cyanhallows7809 Clynes is better known for collecting data on emotional expression than for his compositions. What I took from his work is that music that tries to remain emotionally static runs one set of risks, and music that is more emotionally fluid runs a different set of risks. If the format is two minutes and forty two seconds (20th Century pop format), focusing on maintaining emotional stasis is reasonably practical; failing to do so might even bother a lot of listeners. I see large emotional shifts in pop format mostly in novelty songs or in songs for pretty small niche markets. Conversely, the market for half-hour raaga is a smaller market than for pop songs, and raaga are more or less emotionally static, although, there is effort to intensify the specific emotion. Successful commercial recordings that are anything like 10 minutes long tend to have significant emotional shifts achieved by changes of tempo, meter, scale, lyric content, and melodic figuration. Clynes identifies different optimum iteration rates for figures corresponding to specific emotional expressions. Iterations of grief figures or reverence figures are better further apart than for figures corresponding to joy or anger, for example. It's important to understand that tempo difference is not exactly the same question as iteration rate. At slower tempi, similarly notated repetitions of figure will be further apart. But the same effect could also be achieved by notating them further apart without slowing down the tempo.
Oh damn what a coincidence. I just recently watched a documentary about AlphaGo, the AI that beat the world's top 1 Go player and now David uploaded this!
@@guscox9651 I mean it was a coincidence that David uploaded it on the same day as me watching a video about AI. I wasn't referring to it appearing on my feed suddenly after watching the previous vid since I'm subbed to this channel so it will pretty much appear
This is one of your best videos! Back in the '90s I had a computer program called Fractal Music. You entered mathematical variables and the program would generate an unending musical piece. Fascinating stuff!
As always: Your best video yet. The connex to Numberphile and the many versions of "algorithmic music" while telling a story was masterful. And a good Steve Reich slam never hurts!
This was incredibly well made. Thanks, Bruce! Also, as a percussionist, I REALLY enjoyed your video featuring Gene’s work. Keep it up!!! Will be sharing this
Correct me if I'm wrong, but I think you need a massive ton of data in order to replicate another style, Jacob's catalog, prolific as it is, it's not huge enough to produce clear consistent results
@@whycantiremainanonymous8091 I bet you it's just that we haven't figured out the formula's behind his music yet. Music theory has evolved from people creating music "without a formula", only to have theorists notice things in common between that music (and their contemporaries). Through those trends, they come up with theory to explain how it works. I bet you in 200 years, Collier's music will be just as systematized as the classical period is today.
Wow, amazing video. As someone that's deeply interested in computer science, maths and classical music, this video really let me know more on "Algorithmic music" from a professional composer's perspective.
Analyzing music yields mathematical patterns which we can study, deconstruct and reconstruct to produce new patterns; computers, of course, can analyze and re-create faster and more comprehensively all those possibilities; but ultimately only humans can decide if new musical patterns are at all interesting or pleasing--to humans); and while simple melody seems to have shown the greatest potential for achieving the kind of approval that approaches universality or immortality; on the other hand, complex or more subtle tonalities such as heard in modern orchestral music, polyphonies, electronic--and even "primitive" music and rhythms, may bypass conscious awareness and produce profound psychic and even physiologic changes that are perhaps as, or even more significant to us, organically and holistically as our sentimentally favorite tunes that have "passed the test of time."
Outstanding work and point of view, as always ! It’s really refreshing to see a modern composer not diabolizing algorithms, and reminding us they’ve been around for milleniums. Bravo David !
This guy's exposure is extraordinary ! It's funny as a performer I heard Aki and her brother Yuji Takahashi,Roger Woodward play Xenakis and Yuji and Messiaen,Boulez the once living masters and couldn't appreciate much except that it was very , very different . Then when I began composing and noticed detail , textures, density uncanny , ingenious notational figures etc,Once you try making things (painting,sculpture,architectre , music ) you can really listen and appreciate better !
Great video! Algorithms are frustratingly misunderstood these days and this is one of the most accurate and informative pieces on them from a non computer scientist.
Glad to see you discussing Per Norgard. What I find interesting about his use of the infinity series is that one can apply the basic formula to different types of pitch collections, microtonal ,chromatic or tonal in order to get different results. While Voyage Into the Golden Screen is good as an introduction to his infinity series pieces, I believe it was his first piece to do so and at times it gets too repetitious for me. I find the his second and third symphonies and the harp concerto Gennem Torne more rewarding to listen to.
Thanks for making this video! As a computer science + music major, almost nobody seems to be aware how far along algorithmic composition has come, and how long it's been in development! Although, it's a bit funny to study. At my uni's library, algo music books were either punchcards from MUSIC V "scripts", or David Cope's works, or some isolated paper. Now that AI/ML has hit the mainstream, I can't wait for algo music to evolve (ha) in the next decade or so.
I'm working on a startup to assist composers, and new students. As you say, composition is hard work-computers won't/shouldn't replace humans but they can help education.
I suppose you have been watching a lot of Monty Python... But great video! As always, very interesting. And thank you for introducing Arvo Part to me! P.S: That Xenakis animation is pure nightmare fuel.
I have the notated infinity series signed by Per Nörgård, he was quite confused when I suggested the notation as autograph paper. "But.. I didn't make the infinity series...". But I was happy. Listen to "Gilgamesh" - amazing opera, and of course "I ching"!
Xenakis is extremely effective at getting the kind of result he very clearly intended. That's a reason for other composers to try to understand what Xenakis does and how he does it, even if they don't enjoy the music. Good art is something people probably shouldn't all respond to in the same way. It's pretty hard to make something that seems both personal and universal, because these things are somewhat at odds with each other.
Really great video dude! love the editing and the animations in this one and your other videos lately. Would recommend giving Autechre a go if the idea of algorhitmic (electronic) music interests you.
"Woah, this video is nearly as trippy as Ben Levin's recent stuff..."
*Ben Levin shows up*
*head explodes*
Very inspired by Ben's recent stuff. But I don't think I'm 10% as trippy!
@@DBruce I suspect you're a Monty Python fan as well? I'm getting a strong hint of Terry Gilliam's animations for Flying Circus @ 4:55 e.g..
@@Memento_Mori_Music I came here to say this. In fact I said it, but you beat me to it
This is a pedagogical masterpiece! I knew about like 2 algorithmic approaches before, but now I know about a bunch, but more importantly I am excited to use them in my music! Thank you for your devotion to your art, it’s really inspiring!
I worship at the altar of Benedict the Levine
What do you think of my vowel "solfege" diagram (more like oh-aeh), with only smooth transitions? imanuelhab.mooo.com/scale-major-vowels-imanuel.png
So trippy that I had made this about a week before this video was uploaded! Man, what a perfect timing/coincidental overlapping in time and... pitch!
@@YambamYambam2 what is its origin and use?
I know this comment was a while ago, but is there a link or resource or list of these somewhere?
Huh, even your comments are somehow very Ben Levin-y. To the point of inadvertently reading the comment with your voice!
Editing is like Terry Gilliam mixed with Wario Ware. Weirdly nostalgic.
I knew I have seen this aesthetic before!
"If we change the sound to marimba and start overlapping we quickly get"
A ringtone?
Wow you really surpassed yourself for the editing in this one, love the illustrations !
This might seem weird to say but I think taken to this extent it can be a bit much at times. More quirky and distracting than illustrative.
Rade-Blunner No.
I agree with fikradas
Certainly Monty Python inspired
Brilliant video and great visuals too! Nicely done 🤖
Numberphile and the OEIS are two things I didn't expect to appear in a David Bruce video.
I want more overlap between my maths and music content.
All we need next is a collab that pulls breadtube in, and we've got a real party going on.
I was about to say that!
Can I please get that machine as a guest on my channel? I just love it :) haha
the one with the coin eater face hhh ?
Many musical conventions that have been used for centuries are algorithms in some sense. For instance, the sonata form is a recipe that says something like "Write a theme A and a theme B; Start with the exposition of A, then the exposition of B, then a section of increasing tension that plays with both A and B, then a re-exposition of A, etc.". Doesn't that sound like an algorithm?
Same could be said about many genres in general. It's pretty standard in Blues music to play 12 bar blues along side a pentatonic scale melody, which limits your chord variation right from the beginning. Obviously you can break and bend these rules but they are stylistic components that define a genre. Effectively an algorithm.
From there, the more you define the rules, the easier it is for a computer to create it. I suppose thats why the line between man and machine music is very grey.
omg three of my favorite youtube channels in one video
-numberphile
-ben levin
-david bruce
rock on!
3blue1brown
The Planets AI version killed me 😂😂
This is really inspirational. It both highlights the importance of some sort of algorithmic approach to composition, but also recognises that a more formulaic approach to writing music doesn't necessarily mean that either the music will be banal, or, at the other end of the spectrum, esoteric and impenetrable. You show that with your lush string piece! Also, the editing on this is amazing. It must have taken forever!
Wow that ending was amazing
+1
Indeed!
Very insightful! One of my favourite electronic music composers has recently said that even if they could, they wouldn't share the software they created themselves to help them make music, because even though algorithms definitely play a role, they still need the composer to play them. The software, the algorithms don't do anything without the player. They're all just instruments! It is up to us to make use of them in meaningful ways. And that is ultimately deeply tied to being human.
Which electronic composer was that? Autechre or Aphex? ;)
I feel like I'm watching Monty Python and I love it!
wondering if I can fulfill a 'music theory meets monty python' tagline!?
@@DBruce this video was pretty much it! I'd love if you managed to keep it up, so nice seeing our music theory youtubers stepping up their animation game!
My only complaint is that I spent the whole video waiting for a giant foot to squish someone, and it never came 😪
@@twkotb "and now for something completely different"
Felt the same, love it!
Wow! Someone finally mentioned Norgard! I’ve been mentioning the infinity series to several of the folks who do music theory videos, but no dice😝
This is my favourite video on UA-cam.
Thank you for making this video. I'm nobody important, but I do have a few easy tips which I hope you will consider are just one more reason for people to come here and give you a higher view count:
1) Modern composers now have so many technical resources available that what stands to distinguish them most (as not redundant with other composers) are things like their aesthetic goals and their capacity for meaningful judgment of their options; things that algorithms, alone, still don't provide all that well to listeners. Focusing on clever algorithms will tend to make a composer sound like other composers who focus on clever algorithms. But starting with way too much algorithmic data and gradually paring it down to preferred components in a preferred order will allow the composer's personality and sense of musical value to come through. Is a modern composer cheating by not using pure algorithms? Shhh... I won't tell if you don't.
2) A trick I have used to get around duplicating other people's algorithms is to take readings from a track pad and combine them with pseudorandom numbers under modulo reduction. Just wiggling your finger around on a track pad can really generate a lot of data that's not analytically reducible other than by imposing what are probably false mathermatical patterns.
I'm aware of at least one composer who has used hand-sweeps of a radio telescope across the sky to pick up data. It's unlikely that what happened out there thousands of years ago relates in any discernable way to arbitrary hand motions, so the results should be pretty original in terms of being reliably random. Again, if you're an insightful editor, at least some listeners might appreciate your editorial insight as expressed in musical composition.
3) I don't think Guido's use of vowels was quite optimum. For a language like Latin, Spanish, or Esperanto (and more recently Hindi and Marathi), I prefer the idea of using 2 different (though perhaps related) arrays of 5 pitches; one for stressed syllables and one for unstressed syallables. My preferred order of these vowels in ascending pitch is U,O,A,E,I.
Look at the mechanics of vocalization and maybe you'll see why. The reverse order has some merit, too, if your goal is more narrative than musical.
Similarly, English as phonetically transcribed, has about an octave's worth of vowels. I default shwa to the middle of the range and spread out the others by basically drawing the shortest possible line through them on a vowel space diagram. This should work about the same with languages other than Swedish, at least.
4) The cheapest cheat code to greater melodic interest is simply to increase the proportional frequency of figures that look like "escape tones".
For a complete explanation of why this works, please read (if you dare) The Analysis and Cognition of Basic Melodic Structures by Eugene O Narmour,
and also this thing. www.jstor.org/stable/40285540?seq=1 The short way to combine these things is to bias melodies toward gradual escalation of nonclosure.
Slowly escalating nonclosure basically makes people keep paying passive attention because it's an intensity parameter.
As woodwind exercises to help condition me to improvise better, I have constructed some signatureless diatonic contours that, in a brutally algorithmic way, elaborate an "escape tone" figure with successive elaborations of similar figures; they look vaguely like fractals that expand to the right.
I have no proof that this is helping me, but my subjective impression is that it is helping.
Speaking of algorithms, I've never found the purely random ones like using the digits of PI to be all that interesting as a basis to create music. I definitely dig using highly structured sequences as compositional tools. Norgard's "infinity series" reminds me of another highly ordered set of sequences: superpermutations.
Garry Kasparov's "Deep Thinking" is a great book about AI vs Human relations.
The penultimate chapter is called "human plus machine" that discusses how the intelligence, and perfection of computers PLUS the creativity of humans is the greatest actor we can get. In chess, this manifests in humans making strategies, battle plans, taking enemy weaknesses into account, etc, with the machine giving them perfect predictions on the chances of each strategy succeeding, so the intelligence and creativity can be combined. I imagine that the way that composers might use this is with some kind of lick generator, so for example, if I'm starting on C and need to get to Bb an octave below, the machine could calculate the most satisfying to the ear possible way to get down there in let's say a bar, and from there, the composer could choose weather C-Bb was the best choice at all, or weather this 'perfect' lick could be changed to express something specific. This could also be done with cords, so say if I have two very, non-matching cords the AI could generate a path to that make them fit, so the composer can focus on more important things. This is pretty similar to what many of us do already, I, for instance,have a folder of drum fills or solos so when I have a vague idea of what I want I can look through what I've already made. I think the best place for AI is in this 'human plus machine' dichotomy the same way that no chess machine will ever be able to think about metagame or tilting the opponent, AI will never be able to capture experience, or put into notes nuanced emotions, it can, however, be a great resource for 'checking' yourself, meaning all the calculations a chess player is already making, such as 'are there any holes in my defence' can be answered with 100% accuracy and 0% bias, or in other words, doing the automatic calculations humans already do but without objective error. And composing already has similar automatic calculations, I don't have to think about how to resolve, I automatically default to V - I. Or in other words:
AI is better at the science of music theory, but humans are better at the art of music. and together they'll probably be pretty cool
This already used for a lot of music. Not the one you hear in an opera house, but the one you hear in TV, movies, games and all around you.
Programs like Band in a Box can automatically arrange the melody and harmony input as a Richard Wagner symphony, mariachi band song or Scott Joplin tune. And I'm not saying composers really shoehorn it in like this, they often completely rewrite the output , but they use these tools as part of their creative process, since it automates the boring parts of composing, letting you focus on the more interesting, creative parts...
I'm only halfway through the video, but I had to jump down to write how much I love the editing in this, and your interaction with it! It sprouts with creativity and charm, which shows that you're not really limited by your skills - or rather you perfectly worked within those limits. It's brilliant.
Speaking of composing and inspiration: 1:36 I used to go for walks in the forest with my electric guitar and that exact same little marshall amp on the shelf, just to our left of Davids ear.
Thanks, great topic! I think there are two really-good uses of algorithmic composition:
1. To flesh out details that aren’t super-important. Simplistic example: If you’re writing a 4-part chorale, you write an inspiring melody, and enticing bass line, then let an AI that understands the mundane “rules” of Common-Practice harmony fill in the details, avoiding parallel octaves and such.
2. To build an arsenal of seeds for inspiration, to break logjams in future composition: For example, have an AI write a bunch of melodies, then spend an afternoon auditing them: “crap, nonsense, garbage, ooo wait! that one has some potential; save it away! hooey, rubbish, crap…”
As a software engineer and musician, bless you for covering this topic!!
In the late 80s Steve Coleman who was guest artist at a school I was loosely connected to told me that he'd created a program to improvise jazz based on his own musical predilections. He said it worked well enough for famous bassist composer David Holland to tell him to get rid of it.
David's been upping his editing game during the lockdown!
I love the humor in your videos. Always funny but still very informative and educational! Keep up the great work!
It's worth mentioning that counterpoint has many rules which, while still allowing freedom of the composer, are quite amenable to algorithms. There was a paper where the authors wrote a program to apply the rules of first species counterpoint to show the composer which notes were allowed according to the rules given the notes already chosen.
What is the background music at ~5:40? Sounds great
Visuals were your best yet! It's very inspiring to see you really inhabit a video style of your own
Really fine work David
The animation in this one is very charming. Great work as usual!
Great video. Woujld you be open to sharing that score you created from the Xenakis algorithm? I love the chord colors and I'd like to study them.
I enjoy how much you enjoy editing this Maestro Bruce.
Why only 198000 followers! This is pure gold! Learned so much!
Xenakis is my most favorite composer. I’m so happy that you talked about him. I love all of his works. Thanks for this.
People need to talk about Avro Pärt more, and it’s awesome to see you mentioning him
Fascinating exploration of music composition and algorithms! 🎵 Algorithms offer composers new tools for unique and innovative pieces, generating ideas and shaping composition structures. 🎹 It's crucial to note that algorithms enhance, not replace, creative input. 🚀 As tech advances, the composer-algorithm relationship evolves, opening exciting avenues for musical expression. 🌟
Dear Bruce. This is really an excellent video! Great. I think AI or algorithms can be used to get you to other directions you didn't even thought of. Like you said at the end of the video. I saw once a documentary in German TV about a painter who used AI like a assistant, who gives him new ideas. Looking forward to the next video. Very good and inspiring channel you have. Stay healthy, Michael
I love your thought provoking videos very much, David Bruce. Thank you. I would simply like to add that Guido d' Arezzo original hymn on which he based a new note-naming system was the "Ut Queant Laxis (Hymn to St. John the Baptist)":
Ut (queant laxīs)
re (sonāre fibrīs)
Mī (ra gestōrum)
fa (mulī tuōrum),
Sol (ve pollūtī)
la (biī reātum),
S (āncte) I (ōhannēs) That is: Si.
Later renamed by Cecile Gertken, OSB (1902-2001) as follows:
Do (let our voices)
re (sonate most purely),
mi (racles telling),
fa (r greater than many);
so l (et our tongues be)
la (vish in your praises),
S (aint0 J (ohn the Baptist). Afterwards becoming Ti - in English speaking countries.
As quoted in Wikipedia. org under "Ut queant laxis"
Love the evolution of your video style. Also thank you for the Ben Levin cameo. That guy makes me smile.
Possibly the best video in YT that taps into the zeitgeist? Seriously, well done Mr. David Bruce. This video is a testament
Fascinating! Thank you very much for this.
Really fascinating and quite unknown are two tables by William Bathe (from his "Briefe introduction to the Skill of Song") that can be used to algorithmically write or even improvise canons and canons over a cantus firmus.
Peter Schubert has been using this idea for teaching improvised counterpoint, although I don't believe he has quoted Bathe.
This is an incredible presentation of these ideas! Eternally grateful that you're on YT and not trapped in an academy somewhere.
The Guido vowel thing is something I’m going to put in my back pocket for later, thanks.
thank you David, fascinating !
Love this vid estpecially! Your nerdiest ones are the BEST!
That was quite a horror show.
Very well done. I liked it.
These animations are god tier
Thank you for elaborating on the topic
In my 25+ year programming career, I've lived through (and worked in) several AI fads. They have come and they have gone.
Composers don't have to worry, except that you may be writing less soulless corporate music. The fact is that AI still works best when it is used as an assistant or adjunct to human judgment, and for the foreseeable future, cannot be trusted as replacement for it.
At least until people start treating the work of Manfred Clynes like it's actually important. Once your AI can extract senteme contours from large training sets of commercially successful music, it's probably just a matter of time before there's music that has to be legally controlled because it makes people kill themselves or have sexually compulsive episodes.
@@joshuabroyles7565 Interesting, tell us more - what is the work of Manfred Clynes and is there music possible that could be so condusive to an emotional reaction that it must be controlled? I certainly don't believe so
@@cyanhallows7809 Clynes is better known for collecting data on emotional expression than for his compositions. What I took from his work is that music that tries to remain emotionally static runs one set of risks, and music that is more emotionally fluid runs a different set of risks. If the format is two minutes and forty two seconds (20th Century pop format), focusing on maintaining emotional stasis is reasonably practical; failing to do so might even bother a lot of listeners. I see large emotional shifts in pop format mostly in novelty songs or in songs for pretty small niche markets. Conversely, the market for half-hour raaga is a smaller market than for pop songs, and raaga are more or less emotionally static, although, there is effort to intensify the specific emotion. Successful commercial recordings that are anything like 10 minutes long tend to have significant emotional shifts achieved by changes of tempo, meter, scale, lyric content, and melodic figuration. Clynes identifies different optimum iteration rates for figures corresponding to specific emotional expressions. Iterations of grief figures or reverence figures are better further apart than for figures corresponding to joy or anger, for example. It's important to understand that tempo difference is not exactly the same question as iteration rate. At slower tempi, similarly notated repetitions of figure will be further apart. But the same effect could also be achieved by notating them further apart without slowing down the tempo.
Just wanted to say thank you for your excellent video essays, Bruce. You are my hero.
Oh damn what a coincidence. I just recently watched a documentary about AlphaGo, the AI that beat the world's top 1 Go player and now David uploaded this!
Not a coincidence, the algorithm noticed you were interested in AI so it made sure you watched this one too
@@guscox9651 I mean it was a coincidence that David uploaded it on the same day as me watching a video about AI. I wasn't referring to it appearing on my feed suddenly after watching the previous vid since I'm subbed to this channel so it will pretty much appear
This is one of your best videos! Back in the '90s I had a computer program called Fractal Music. You entered mathematical variables and the program would generate an unending musical piece. Fascinating stuff!
As always: Your best video yet. The connex to Numberphile and the many versions of "algorithmic music" while telling a story was masterful. And a good Steve Reich slam never hurts!
This was incredibly well made. Thanks, Bruce! Also, as a percussionist, I REALLY enjoyed your video featuring Gene’s work. Keep it up!!! Will be sharing this
What would happen if we threw in a bunch of Jacob Collier's music into one of those computer algorithms and see what they come up with?
that would be fun to see!
Actually, it probably would sound like nonsense. Collier's music is not especially formulaic or predictable.
Correct me if I'm wrong, but I think you need a massive ton of data in order to replicate another style, Jacob's catalog, prolific as it is, it's not huge enough to produce clear consistent results
@@whycantiremainanonymous8091 I bet you it's just that we haven't figured out the formula's behind his music yet. Music theory has evolved from people creating music "without a formula", only to have theorists notice things in common between that music (and their contemporaries). Through those trends, they come up with theory to explain how it works.
I bet you in 200 years, Collier's music will be just as systematized as the classical period is today.
WhyCan'tIRemainAnonymous?! I disagree
Wow, amazing video. As someone that's deeply interested in computer science, maths and classical music, this video really let me know more on "Algorithmic music" from a professional composer's perspective.
Analyzing music yields mathematical patterns which we can study, deconstruct and reconstruct to produce new patterns; computers, of course, can analyze and re-create faster and more comprehensively all those possibilities; but ultimately only humans can decide if new musical patterns are at all interesting or pleasing--to humans); and while simple melody seems to have shown the greatest potential for achieving the kind of approval that approaches universality or immortality; on the other hand, complex or more subtle tonalities such as heard in modern orchestral music, polyphonies, electronic--and even "primitive" music and rhythms, may bypass conscious awareness and produce profound psychic and even physiologic changes that are perhaps as, or even more significant to us, organically and holistically as our sentimentally favorite tunes that have "passed the test of time."
Outstanding work and point of view, as always ! It’s really refreshing to see a modern composer not diabolizing algorithms, and reminding us they’ve been around for milleniums. Bravo David !
the editing recently has been incredible
Came here from tantacrul video about corporate music. Lovely channel.
Stay safe dude.
I like the new style
This guy's exposure is extraordinary ! It's funny as a performer I heard Aki and her brother Yuji Takahashi,Roger Woodward play Xenakis and Yuji and Messiaen,Boulez the once living masters and couldn't appreciate much except that it was very , very different . Then when I began composing and noticed detail , textures, density uncanny , ingenious notational figures etc,Once you try making things (painting,sculpture,architectre , music ) you can really listen and appreciate better !
Love the Gilliam-esqe...ness of your videos. Really adds an extra level to the already impressive depth of research and accessible presentation.
Great video! Algorithms are frustratingly misunderstood these days and this is one of the most accurate and informative pieces on them from a non computer scientist.
Glad to see you discussing Per Norgard. What I find interesting about his use of the infinity series is that one can apply the basic formula to different types of pitch collections, microtonal ,chromatic or tonal in order to get different results. While Voyage Into the Golden Screen is good as an introduction to his infinity series pieces, I believe it was his first piece to do so and at times it gets too repetitious for me. I find the his second and third symphonies and the harp concerto Gennem Torne more rewarding to listen to.
Thanks for making this video! As a computer science + music major, almost nobody seems to be aware how far along algorithmic composition has come, and how long it's been in development!
Although, it's a bit funny to study. At my uni's library, algo music books were either punchcards from MUSIC V "scripts", or David Cope's works, or some isolated paper. Now that AI/ML has hit the mainstream, I can't wait for algo music to evolve (ha) in the next decade or so.
I'm working on a startup to assist composers, and new students. As you say, composition is hard work-computers won't/shouldn't replace humans but they can help education.
Oohh Arvo Pärt :) Very happy you mentioned him! Long live rule-based music.
omg, a david bruce upload and Tantacrul in the same day!! Today is a good day
Another brilliant video, now watching for the second time! :D
loved the meme style editing in this video
I suppose you have been watching a lot of Monty Python... But great video! As always, very interesting. And thank you for introducing Arvo Part to me!
P.S: That Xenakis animation is pure nightmare fuel.
It would be awesome to have a 2nd part of this video! This was great! Thank you 🙏🏼
Wonderful, finally an unbiased look at algorithmic music, nicely done 👍
I have the notated infinity series signed by Per Nörgård, he was quite confused when I suggested the notation as autograph paper. "But.. I didn't make the infinity series...". But I was happy. Listen to "Gilgamesh" - amazing opera, and of course "I ching"!
The editing in this video is so good!
It was pretty cool to see Ben doing music in that style.
U actually talk about xenakis one of my fav composers. noice.
Xenakis is extremely effective at getting the kind of result he very clearly intended. That's a reason for other composers to try to understand what Xenakis does and how he does it, even if they don't enjoy the music. Good art is something people probably shouldn't all respond to in the same way. It's pretty hard to make something that seems both personal and universal, because these things are somewhat at odds with each other.
just heard some of your stuff on R3, David... great! Also love this film...especially the 'plate spinning' DJ
Really great video dude! love the editing and the animations in this one and your other videos lately. Would recommend giving Autechre a go if the idea of algorhitmic (electronic) music interests you.
All hail Benedict the Levine, God bless you !
Love it! Working with limitations set creativity free
I didn't know I needed to hear ben Levin singing Latin until today
thank gosh i have such an awful sleep schedule so i can watch this
Lol get rekt it’s 1 pm here in England lol
Absolutely brilliant video, even for your standarts. Bravo
Loved the Ben Levin cameo!
“Good ideas aren’t made by algorithms, they arrive mystically through the ether” I died with David’s face 😂
ua-cam.com/video/TZMmlN1YtyA/v-deo.html
@@joshuabroyles7565 hahaha
Love the Monty Python style of editing. Great video !
Quite interesting - thought provoking, actually. Thanks so much for posting this.
Arvo Pärt is an absolute genius!
This editing is amazing
Amazing visuals on this one!
Really excellent video, thanks for all the inspiration & education :)
Whooooa, that is really beautiful at the end. You gonna make an entire composition from that, David?
Yes, would love to hear that! Also to know how you used Xenakis' data set to produce it.
Could you describe the process that you used to generate the chords from the Xenakis data?
David, your Xenakis-algorithm string chords are gorgeous! I was trying to figure out what that piece was when it first came up! :)
Thanks for a great intro to this fascinating topic.