Leonard Bernstein did an excellent Omnibus talk in the 50s where he reconstructed Beethoven's numerous discarded sketches in the 5th symphony (stating that the 2nd movement theme went through at least 14 iterations), and demonstrating exactly how Beethoven's final product was superior and the struggle Beethoven went through to create the work we know and love. Any knowledge of this 'creative process' should put to rest any attempt to recreate a Beethoven symphony using nothing but a series of preliminary sketches.
@@steveschwartz8944 One thing I loved about his demo was that 90% of composers--the ones not LvB-- would have been satisfied with ones of the alternatives and let it go at that. And have been accounted very good composers yet ones who evaded greatness. Yes, a great teaching strategy.
What a thorough rebuttal against "stupid"...or is it ignorance? This reminds me of the ancient Taoist saying, "Those who know, do not speak. Those who speak do not know." But in this case, I'm glad you are speaking. I'm a survivor of academia (having gotten a doctorate in music and gone on to be a "professor" of music). This Beethoven's 10th project reeks of the usual suspects: the good ol' boy's club I was fortunate to escape from. Thanks for this brilliant presentation. My hat (and scarf) are off to you, sir!
I'd like to add that another way of putting my response to your "push-back", David, is that it's probably the best "elephant in the room" discourse ever.
As someone who has a fair bit of experience with AI, many of these computer scientists and media outlets are dishonest in describing their AI compositions. AI and machine learning, as it exists currently, just isn't very good at writing large-scale sonata form in the way that an actual composer would, so, oftentimes, these AI compositions consist of small AI-generated fragments which are modified and arranged by modern composers after the fact. You create some sort of markov chain to predict the next interval/rhythm given a start, and the markov chain generates some basic melodies and harmonies. Idk if that's specifically what happened here, but oftentimes these "AI-generated" compositions are just a way to legitimize compositions. People think that, because a composition was computer-generated, it is somehow more true to the original beethoven than a work completed by a human composer. (This is obviously nonsense; AI has many biases of its own, and purely AI-generated compositions very clearly fail to grasp musical nuances apparent to human composers.) Beyond that, AI-generated compositions tend to attract more media attention, so the computer scientists/composers involved in them tend to overstate the role of AI/machine-learning in generating these compositions. It's practically a cottage industry at this point, and your skepticism is entirely warranted.
Agree, at its mathematical best the most perfect outcome would be "average". This is a simple fact of how this technology works. The miracle of human achievement is its variability and creativity - which is not part of these procedures and WHY we listen to Beethoven in the first place.
It reminds me of a colleague in an introductory AI course. The program she wrote was to bake the Perfect Chocolate Cake. She basically had a fixed list of ingredients and would bake a chocolate cake, submit it to a taste test, and fiddle with proportions for the next round, based on comments she received. She wound up with a cake that tasted like a Duncan Hines cake mix. She eliminated me early from the tastings because I was always the outlier. I preferred a European torte. I don't like American cake mixes.
I would like to express a disagreement. Not all algorithms rely on auto-generated music in Markov, especially with the naïve way that many times referred to. Of course, there are a Great number of improvements that have to be done to model the musical creativity, but It is oversimplified to say that "it is just a Markov model". Long-Short memory, automatic GTTM analysis, and many others can address the problem of local probability dependence. Politely!
Thank you for restoring reason! When the austrian composer Gottfried von Einem was asked to finish Bruckner's Ninth, he said no to the project, but used the material for a work he called "Bruckner-Dialog". Luciano Berio did a similar thing with Schubert in "Rendering". I guess, that's a creative way for dealing with fragments, sometimes with more, sometimes with less fortune. But to finish a work, which, in fact, never was really started, and calling it, at least in the subtext of labeling it "10th symphony", "BY Beethoven" is, as you say, hybris. It's one of the greatest chuzpah, I ever heard of.
The mind boggles at what AI could give us. I've always wanted to hear the Délibes Banjo Concerto. I believe there are no surviving sketches, but hey, that's no problem.
As an engineering and computer science major, I find the idea of "finishing" an unfinished work to be completely absurd in general, but especially with AI. I'm curious to know if they even tested their algorithms on other Beethoven works. The only way to get an idea of accuracy would be to feed their program fragments of a known work of Beethoven's, and then see if what it spits out is close to the real thing. Forget all the intricacies orchestration, I wonder if this thing could even accurately reconstruct one of his Piano sonatas (which are still enormously complex works).
Someone once pointed out to me that "Logic Omits Us" is an anagram for "musicologist". Coincidence? I think not. Thank you for voicing out some reason.
P.G. Wodehouse (author of the Jeeves books) left us a clear account of his creative process: "I just sit at my typewriter and curse a bit". Beethoven's creative process was probably the same, except that he didn't use a typewriter, and he cursed more than a bit. I doubt AI can use that process to give us Beethoven's 10th, any more than it can give us "Jeeves and the Pandemic Shutdown".
Perhaps a century from now, cyber archeologists will unearth the Ask Jeeves search engine and use it to recreate all the Netscape-designed web pages that populated GeoCities.
I read the article you critique a day ago myself, and your deconstruction (or should that be decomposition?) of this more than slightly pretentious and ultimately sterile project is spot on. From musical juvenilia such as his completed-but-surviving-only-in-fragments Oboe Concerto in F Major to the barely begun String Quintet in C Major he was working on at the end of his days, we can only mourn what we have lost of Beethoven's music, and what he might have written during a longer life. Far more importantly, however, we can be infinitely grateful that such a genius ever existed at all, and that he gifted humanity with so many masterpieces that we do, and always will, have and cherish. On a personal note, I just realized that this is essentially the 25th anniversary of one of my earliest published (in the magazine Analog Science Fiction and Fact, October 1996 issue) science fiction stories, "Symphony in a Minor Key." It is a combination time travel-alternate history tale in which an admiring musicologist from the future changes the past and gives Beethoven seven additional years of healthy life. Naturally, one of the revived composer's creations is the Tenth Symphony, although in this imaginary tale it is an entirely de novo work and not based on the E-flat Major--C Minor sketches. For obvious reasons I could only describe the barest details of what the symphony was like and how it sounded. My novella focuses instead on what impact the Tenth had on the subsequent history of music and the world. (Spoiler alert--to put it mildly, things do not turn out well...)
The real hubris, as you point out so entertainingly, is the assumption that you could capture Beethoven's creative process at all, even from the entirety of his surviving work. Is that what a creative process is, the sum of parts, something you can put in a box? I don't think you can separate out anyone's creative process from external conditions and circumstances, even something as trivial as what you had for breakfast, or what the weather is at the moment, or reading Goethe at bedtime. I agree that the project is fascinating, but it cannot justify claims to being Beethoven. Sheesh!
It's a question of marketing. AI Nerds: We taught the AI to take some fragmentary Beethoven sketches and turn them into something that sounds vaguely like Beethoven. Musical Public: Good for you, I guess. AI Nerds: We taught the AI to write Beethoven's 10th Symphony! Musical Public: Wow! Let's hear that!
Karl Holz, violinist of Schuppanzigh quartet and a good friend of Beethoven at his last years, claimed that he played the first movement on the piano. Prior to his death, at March 18, 1827, Beethoven wrote to Ignaz Moscheles that he "is still hoping to thank the London Philharmonic Society for costless help and to dedicate his new symphony which sketches lie on his table, or the new overture, or something else".
From the way the article is written (like explaining what is a fugue and a coda, what does orchestration mean, and using cliched phrases like "taking the baton from Beethoven"), it seems that the 10th symphony was composed to please the non-musicians... like "Ah, you like the ode to joy, here's the 10 symphony composed by AI."
It's hard to think of a more unpredictable composer in the symphonic form. He didn't follow a beaten track, he created a whole new road. As you say, who could possibly predict what came next? And then of course there's that tricky thing called genius. Interesting to hear that AI was getting better at that! Thanks David for this hugely enjoyable video. Great to hear your lethal, articulate, humorous scorn put to such great use, dismantling this this absurd enterprise. Great stuff.
I don’t understand why they don’t call it something but like “Symphonic Fantasia on sketches of Beethoven.” Clearly if 99% of it wasn’t written by Beethoven it can’t be said to be by him, or to merit a place within his numbered symphonies.
Very interesting point about a Beethoven being an evolutionary composer. Maybe you'll do a video on evolutionary composers vs. non-evolutionary composers.
Oh dear, I read about this. And I heard the excerpt. It sounds like a bunch of Beethovenian themes, lifted from his other works, strung together into a jumble which goes absolutely nowhere. As an AI experiment, it's interesting. As a musical experience, it's pointless, IMO.
David, I don’t have the time to listen to all of your video at the moment, but your title alone made me spit out my morning coffee from a laugh! Thank you, I so needed that laugh.
Nice one Dave - there's a bigger point about AI and creativity but applied here you have nailed it. Much better than bothering with any 10th is spending your time with the 7th or the 4th or the 2nd ............etc
Love your videos David. They’re extremely enlightening and just so much fun. I was wondering: could you do a video on Beethoven’s Missa Solemnis? I couldn’t find a video on your channel. Hoping you tackle it soon!
They started shooting "Casablanca" before they had a completed script, and the actors didn't know how the story was going to end. Ingrid Bergman didn't know which of the two leading men she was supposed to be in love with. An AI-generated completion would have had her run off with Bogart, because he was the bigger star, and that was how Hollywood usually decided things. It would have ruined the film overall, but it fit the usual formula, and that's what AI prioritizes.
Maybe we could get an Eighth and Ninth Symphony of Sibelius and Prokofiev in the process, the 28th piano concerto of Mozart, a sixteenth string quartet of Shostakovich... Beethoven broke the rules and got away with it, and that is the mark of true genius.
At the present state of AI, I would expect such a project to produce a few passages or perhaps a short movement that sounds "in the style of Beethoven". But the notion that it could start with a few sketches and generate something resembling what Beethoven would have written had he survived long enough to complete Symphony no. 10 is difficult to take seriously. A test of how close the AI could come would be to give it similarly brief sketches from the composition of one of the completed symphonies and then see whether it produces a piece resembling the real thing (after first deleting the completed version of that symphony from its database to prevent cheating!) For some reason, this reminded me of the efforts of Rosemary Brown to generate new compositions by dead composers while acting as a spiritual medium. I think that was back in the 1970s. I only remember her name because I knew someone else with the same name and sometimes asked her how the composing was going.
Could AI have created THE NINTH or Quartets 13 & 14? Or the last three piano sonatas if they had all the works written earlier before they had existed? I wrote this at the start of the video and see you covered these points later. While I’m at it, how about an AI shot at completing Turandot or the Mozart Requiem and compare with what we have now?
If the experimenters had possesed just an ounce of reason, they would, before embarking on "the 10th", have started out feeding their computer all information pertaining to Beethoven's symphonies 1 through 8, their (alleged) information about Beethoven's composition processes, and plenty of thematical material from symphony no.9 - and then taken a look at, whether their IA was able to come up with anything that was just close to the actual 9th. I can't imagine that happening - judging from the sound snippet of "the 10th" in the page, that David links to, I'm actually certain, they wouldn't by any standard get even close. It sounds like a hotch pot of bits and pieces from earlier Beethoven ideas, albeit structurally and developmentally awfully realized and in such an amateurish manner that it would have sent good ole Ludwig fuming with rage or brought him to the utmost brink of despair!
@@jensguldalrasmussen6446 Exactly my argument. And that scherzo, as I said, was unspeakably banal. Science and experiment demand controls. Giving AI the sketches of an existing work and seeing what it came up with would prove that the claims made for the Tenth are ludicrous. Though as Dave says, experimentation with AI in itself is just fine--but without such gimmicks.
One of your best, ever, chats. Not one of the "great minds" putting this together, could, with any certainty, predict the next 3 minutes of their own lives, based on the sum total of their own past; but are telling us they could predict what Beethoven would have composed, based on the sum total of his output.
Leonard Bernstein once said that Beethoven’s greatest strength was his sense of form: everything that happens in a Beethoven symphony happens at exactly the right time, is the only thing that could possibly happen at that point (or something like that). I listened to the excerpt that Dave included a link to and found it to be a complete mess and muddle in terms of balance and form - totally lacking the quality of inevitability that Bernstein noticed and praised in Beethoven’s writing.
I've just come across a new work - we should be grateful it was headed up as 'new' I suppose - Richard Wagner in Venice, composed / realized / imagined by composer Mathew King where he has supposedly knitted together fragments of orchestral works that had been begun by Wagner. Just let them rest!!!
Great video and wow! I never knew people were attempting this sort of thing. This is what I believe in regards to this subject: only Beethoven can write Beethoven... and he's dead...! so there's no more Beethoven to be had. As much as I love him, its as you said... we've got enough of his stuff. I loved how you said "the ninth" with your previous video on "the ninth"!! :-)
A musicologist named Gerd Prengel constructed movements from Beethoven's sketches that I believe was the most "Beethoven" sounding piece that was not actually written by the composer. (Especially the "Scherzo/Presto".... Although it does have a similar character to the second movement of his Ninth.) He has these compositions posted on UA-cam. I don't think he used AI to construct these movements.
I agree with all you say here Dave, including the respect for trying things with AI. Also think, as this is a general issue raised by this channel and the other review magazines/ sites, there's the question of how much is enough?! These completions (Levin more recently and others) are particularly noticeable in Mozart. Every anniversary we get more and more little works and fragments completed or not. Didn't we have enough already? I mean there's tons of completed Mozart and Beethoven that if I am honest, I can hardly find the time to be always working my way through and listening to! There are exceptions that prove the rule (Berg's Lulu is definitely one, with of course vastly more basis in what was sketched in short score). But wouldn't it be better is we focused more on writing new music? Or even getting machines to write new music to hear what happens?
Your video prompted me to give a listen to the two movements, they are up on UA-cam. They sounded to me like a grad school homework assignment--here are some Beethoven sketches, write two movements of music in Beethoven's style. The AI's achievement was astounding for AI, fairly mediocre for a human who might have tried the same thing. I thought the movements had a lot of oddly dissociated fragments and strange direct quotations (5th symphony and Pathetique Sonara) that Beethoven would never have done in a million years. And why all the organ music in the AI finale? Is there some historical evidence for this? It sounded just weird to me. I think the movements are an impressive achievement, but only if billed as "An AI Fantasy on Sketches by Beethoven."
I once read the complete poems of Yeats. Since he seemed to write in clumps of poems, every so often I'd say to myself, "Oh, I see how he's doing this. It's a matter of certain key words and narrative strategies." Then he'd do, in Monty Python's phrase, Something Completely Different. The excerpt was interesting, but not particularly Beethovenian. The closest it came was certain harmonic progressions in section A. Even here, however, it sounded like pastiches of other works. It became a game of Name That Beethoven. The B section had the same problem, but the material and argument were hapless, sometimes ridiculously so. Nevertheless, as David points out, to me it's quite a feat to get the computer to do make even minimal musical sense.
We’re all entitled to her opinion and a number of good points have been made. But I think the effort is still worth it even if what comes out it’s not quite Beethoven. We have to be inspired to do the impossible.
Fot the sake of verification they should do this the other way around. Reduce the Ninth to the most basic themes, more or less to the same level as the supposed "sketches for the 10th", give it to the AI, and see if it comes up with anyhting remotely similiar to the actual Beethoven 9th.
It might be interesting, as a test of the claims for this AI methodology, to submit to this team preliminary sketches of a completed Beethoven symphony, say the "Pastoral", gleaned from whatever musicological libraries have them in their collection. Let them ply their digital magic and then compare the result to the actual 6th. I'd like to keep an open mind, but I suspect that the difference in form and quality would be substantial. And "two movements of more than 20 minutes"??? Are they sure that someone didn't mischievously slip the dreaded Bruckner virus into the software?
Excellent chat indeed, with which I am in total agreement. I'm sure Beethoven abandoned many more sketches than he ever used (Remember Bernstein's enlightening and entertaining TV lecture on the rejected sketches for the Fifth?). The arrogance of the people involved in this project is beyond belief. At least Deryck Cooke, when, with much much more tangible musical material to go on, he presented his Mahler Tenth projects, had the modesty to call them "performing versions." As for anyone craving more Beethoven, I admit I sometimes do--but middle-period Beethoven. As valuable as Fidelio is, considering all the time, effort, squabbling, revising and rewriting that went into it, I often think I would trade it for a couple more Beethoven symphonies, concertos or sonatas.
Are you kidding me? The guy who wrote the Intel sustained note is the composer who got that gig? If you want something novel and faithful you'd have a better chance putting it into the hands of heavy metal and acid rock musicians who are steeped in the Bach Rock tradition, starting with the late Frank Zappa.
Feed the AI everything about Beethoven as they did, teach it everything, then feed it ALL the surviving sketches of any Beethoven symphony, but not the symphony itself. Watch it produce NOTHING like the actual symphony that exists, I guarantee it. I just heard the scherzo sample from the link, it's unspeakably banal.
I've said elsewhere, if people want to feed some Beethoven sketches into a neural net and record whatever it churns out, I have no problem with that and in fact I find it kind of interesting. But my god, don't be so pompously daft as to call it Beethoven's Tenth Symphony!
Love your comments. Even if there were more sketches left about Beethoven's first conception of a tenth symphony, no one, not even an AI, could estimate how Beethoven would have shaped the final work. All his late work, sonatas, string quartets, the 9th, late variations, even smaller works like the Bagatellen, contains formal experiments that make his progress in future projects completely unpredictable. It's just a technocratic stunt, but not a serious scientific musicologal progress.
Goddamn be those PR people. Because it would perfectly interesting to listen to the “Symphony in Beethoven style” by BOThoven 1.0 as a technological experiment. Because that’s the interest: a machine can emulate someone’s style. Emphasis on machine. Because any mediocre student could have composed that faux-Beethoven scherzo. The Chinese phone manufacturer Huawei once created an A.I. to finishes Schubert’s “Unfinished”. They even hired Alondra de la Parra to conduct its premiere. It made a big splash on tech circles but musically it was ridiculous: the completion sounded nothing like Schubert and absolutely nothing like the preceding movements. In fact, it’s so happy and cheesy that it would be perfect as the soundtrack for some “child-dog movie”.
Here's a suggestion which, if followed up, would keep the ASFI experts hurtling round the world for years to come. Feed a computer with every orchestral work by Beethoven from the second symphony onwards. Then get it to reconstruct the first symphony. Then compare and contrast. The possibilities, if unimpeded by common sense, are endless, and I look forward keenly to Hildegard of Bingen''s saxophone quartet, and Josquin's Missa "Sophisticated Lady". ASFI stands for "Artificial Substitute for Intelligence".
To participate in what sounds like a fabulous junket, I would not hesitate to help complete not only Beethoven's 10th symphony, but also his 11th and 12th!
Teaching the AI about Beethoven's musical processes isn't enough. To get convincing fake Beethoven you have to teach the AI to be German, deaf and cantankerous, live in Vienna in the 1820s, simultaneously admire and strive to surpass Haydn and Mozart, pursue unsuccessful love affairs with a string of women, contemplate suicide, sue for custody of it's nephew (first teaching it what a 'nephew' is), etc., etc., etc.
@@DavesClassicalGuide Interestingly, according to his secretary Anton Schindler, Beethoven washed his hands quite frequently, but he made no mention whatever about the master cleansing the rest of his body. Must have gotten pretty ripe, especially in warm weather. Yet another odd variable for the AI to ponder.
This is so silly. Beethoven left vast swathes of music. The quantity and depth of what's already there are enough for a couple lifetimes. I doubt the people on this project have fully contained the scope of his output either! Seriously nothing more is needed lol
What's next the 'realisation' of Symphony no.0? As there are sketches for a C major symphony that Beethoven wrote before working on his first symphony, these are not related to the first symphony. So basically this piece of music should really be called 'Symphonic Fantasia on Fragmentary Sketches of Beethoven'.
What I don't get is why the AI people took a detour to go on bended knee in front of the great man theory of holy german art; Deep Blue didn't become famous for accurately mimicking Kasparov, it became famous for BEATING Kasparov.
This entire project is nothing but a cheap publicity stunt for Deutsche Telekom, the former state-run corporation that single-handedly made us Germans have one of the worst internet infrastructures in the entire Western hemisphere. I like the Barry Cooper movement sort of. I don't really hear it as Beethoven's 10th, but (considering the form) as a quite original and enjoyable standalone. The 19th century didn't produce something quite like it in its formal concept (although a vast number of much better pieces of course), all the more reason why I find it weirdly refreshing. Probably something you could call a guilty pleasure.
The sample I heard is actually reminiscent of Mozart's musical joke only much, much more hilarious (simply because the music has no sense to it whatsoever).
even if they are just seeing what ai can do, do these people realise the sickening affront to art and human expression these recreations are? i find it amusing
I suppose whenever you mention completions of "The Tenth" you should show up in a Speedo rather than a tie. I think we'd rather forego the results in both cases. lol.
When you actually listen to the little excerpt, you will notice that it is just a combination of symphonies 5 and 9. Did the authors of this project not listen to their own creation? Pardon me, Beethoven's creation. Do they consider that Beethoven created new things and evolved without repeating himself every second or so! David's talk was entertaining but the music itself was hilarious.
You can bet that all of these artificially intelligent musicologists will be living it up with the foundation paying the bill. Nothing that is musically artful will result, though. It will be artificial, but not intelligent.
I'm one of those old curmudgeons who thinks unfinished works should be left alone. Finishing an unfinished (or un-begun) symphony might be a fascinating AI experiment, but I think humanity has more pressing matters for which to utilize that technology.
Speaking of curmudgeons, imagine an AI program that was designed to write "new" columns, books, and texts by Oscar Wilde, James Thurber, Mark Twain, H.L. Mencken, and Sinclair Lewis. Why, that sounds like the plot of a Roald Dahl short story.
Does a Beethoven sketch written by Beethoven sound like Beethoven? Wellll....full stop. If the sketch came from the mind of Beethoven it is indeed Beethoven. Having said that, a cluster of sketches, even if Ludwig intended to use every note does not make it a symphony.
AI is killing the arts business, I put it that simple, computers are not human and humans are not computers, can a computer feel emotion for MUSIC? I admit that it is more fun and even tender to see robots conducting orchestras as in the case of Yumi or Asimo, but how do you teach a robot to appreciate Music? How do you force a computer to think or feel like humans? For a reason it is called "Artificial Intelligence". Let us remember that Beethoven had a long time wanting to adapt Schiller's Ode to Joy (Schubert also made a lieder on the text), And if for something the last quartets have more than 4 movements and the 9th is extensive, it is because I think that Beethoven was testing all his compositional power before leaving something unfinished at the time of death, he already wanted the Ode to Joy to be as a separate chorus, that is why he had the desire to include this chrous and extend Schiller's text in a heavenly way before leaving the world, because he knew that it was going to be his last symphony, who knows how the 10th would have been and towards what directions Western Music would have taken, who knows if it would had a number of bars close to a Mahler symphony, is the same case of Schubert's unfinished symphony, maybe Schubert didn't want it to be performed in public, but I think the fact that it's incomplete is what makes it interesting, In any case, Schubert's style would not have approached so much that of Bruckner (and that both had Simon Sechter as a teacher of counterpoint), and now the AI wants to complete Schubert's Unfinished, Bruckner's 9th or Mahler's 10th and maybe (although I don't want it to happen) Sibelius's 8th. As if human talent is no longer enough for the Arts or entertainment industries that great works are no longer needed, another demonstration that human beings do not care about legacy and preservation, for lack of instruction and respect for Culture.
But seriously, folks, why do we do this with composers... and authors (Tolkien comes to mind)? Here is my humble rumination written elsewhere on the idea of musical completions, in this case the work being Mozart's Requiem: "By keeping alive the puzzle of Mozart’s deathbed creation, we are in a sense keeping Mozart himself alive, denying to ourselves that his magnificent opus is complete. We are in essence keeping guard over Mozart in a perpetual wake, refusing to bury 'the miracle that God allowed to be born in Salzburg' and allowed to die too young in Vienna."
Only two reactions possible to the horrible daft enterprise - and I appologize beforehand to the poor cleaning lady (no matter of which sex): 🤮 and 😭 (unfortunately I couldn't find one emoji, that combined the two of them!). Why bother with artificial intelligence, when you do not posses the genuine article? The whole business reminds one of a sick joke - and, indeed, I had to pinch my arm and make sure I hadn't been somehow teleported back in time to the first day of the month of April. The stilted, pompous baloney, that was read from the commentary, immediately brought to mind the fairy tale of my famous compatriote: "The emperor marched in the procession under the beautiful canopy, and all who saw him in the street or out of the windows exclaimed: "Indeed, the emperor's new clothes are incomparable! What a long train he has! How well they suit him!" Nobody wanted to let others know, that they saw nothing, as they then might have been deemed unfit for office or too stupid. Never before had the emperor's clothes been so profusely admired. "But he has nothing on at all" a small boy suddenly exclaimed. Oh, thou ever young curmudgeon, lift up thy voice, reveal their empty posturing, expose them, bare-assed, for all to see! And we will keep on listening!
I'm in agreement with all who here think this is simply just wrong. 1st this is simple...These are men trying to make a name for themselves and this is entirely self motivated. Beethoven or any composer for that matter has the right to keep their music strictly theirs and theirs alone. They share and we listen period. I think this is ridiculous. the greater point is not Beethoven or this music. It's the status hungry that is the real point. No matter what they do it will be wrong. It is and will never be Beethoven's from his head period. arrogance and glorifying themselves. ALso, in closing how utterly disrespectful to the man's heart. Aside from his Genius and all he went through to write these pieces. As a composer myself. I believe they call these pieces because they hold such precious pieces of ourselves that we share with others and to think not only that YOU are capable of taking that and also giving it to a computer? Total rubbish! let them pretend to be something all they want. we can just shake our heads and quietly walk away ...they will have to face the man some day .....so let them live with that alone in the dark when no one is looking.....but him :) hope this was ok to say....but its offensive to me.
In my view, the only way the musical sample at the end of the article might merit any comment is by assuming it was meant as a parody on some of Beethoven's music. But it fails even under that assumption, because it's not funny or witty at all. It's just embarrassing. My message to the distinguished (?) team of experts: Go get a life!
LvB would have never ever used THE EXACT rhythmic motif as in previous symphonies, let alone the augmented one in 3rd mvt of the 5th Sym. (augmentation of the motiv from the 1st mvt). This attempt has traces of Mr. Levin's grandiose narcisism- it is obvious that he chose the computated options ala- it could have been this, it could have been that. The attempt uses motifs from various B' works in a 2nd grade fashion. This is precisy how a composer's mind DOES NOT work. He/she invents new on the basis of previous experiences-just look at B's 32 piano sonatas. No serious composer would have ventured into this. It is not good, it is way less than a 10th would have been. Go play with some 2nd grade sketches from elsewhere, omg
Having listened to the excerpt, I'm already bored. The music simply doesn't go anywhere, it doesn't progress harmonically or thematically. It's just note-spinning. An insult to Beethoven, if you ask me, but I don't suppose Beethoven would have minded. Fortunately, he wouldn't have been able to hear it anyway.....
Beethoven's 10th would have looked forward toward unexplored vistas as did the late string quartets and late piano sonatas. This AI version is stuck producing mediocre versions of much earlier Beethoven. Garbage, and should be illegal.
I am in the minority who think Beethoven should've stopped after composing his 7th Symphony. His Ninth fell short of the perfection of the 7th, and should have been replaced his 6th, an entirely superfluous composition. I'm not being serious here, except I'm completely serious. Pick one.
Leonard Bernstein did an excellent Omnibus talk in the 50s where he reconstructed Beethoven's numerous discarded sketches in the 5th symphony (stating that the 2nd movement theme went through at least 14 iterations), and demonstrating exactly how Beethoven's final product was superior and the struggle Beethoven went through to create the work we know and love. Any knowledge of this 'creative process' should put to rest any attempt to recreate a Beethoven symphony using nothing but a series of preliminary sketches.
Oh yesterday remember that. excellent
Your commitment to the 'the Ninth' bit is commendable.
We should feed Beethoven’s letters to the AI so he can write his own program notes!
This was hilarious. Even many of Beethoven's sketches or thematic scribbles for completed works didn't make it into the final score.
Good one!
@@steveschwartz8944 As Bernstein demonstrated with the Fifth.
Yes! That's exactly what I was thinking of. One of the great musical lectures of my life. What a brilliant teaching strategy it was, too.
@@steveschwartz8944 One thing I loved about his demo was that 90% of composers--the ones not LvB-- would have been satisfied with ones of the alternatives and let it go at that. And have been accounted very good composers yet ones who evaded greatness. Yes, a great teaching strategy.
I read the article and immediately went to youtube thinking "I hope Dave is doing a video on this". Lo and behold it was the first one recommended.
What a thorough rebuttal against "stupid"...or is it ignorance? This reminds me of the ancient Taoist saying, "Those who know, do not speak. Those who speak do not know." But in this case, I'm glad you are speaking. I'm a survivor of academia (having gotten a doctorate in music and gone on to be a "professor" of music). This Beethoven's 10th project reeks of the usual suspects: the good ol' boy's club I was fortunate to escape from. Thanks for this brilliant presentation. My hat (and scarf) are off to you, sir!
I'd like to add that another way of putting my response to your "push-back", David, is that it's probably the best "elephant in the room" discourse ever.
As someone who has a fair bit of experience with AI, many of these computer scientists and media outlets are dishonest in describing their AI compositions. AI and machine learning, as it exists currently, just isn't very good at writing large-scale sonata form in the way that an actual composer would, so, oftentimes, these AI compositions consist of small AI-generated fragments which are modified and arranged by modern composers after the fact. You create some sort of markov chain to predict the next interval/rhythm given a start, and the markov chain generates some basic melodies and harmonies.
Idk if that's specifically what happened here, but oftentimes these "AI-generated" compositions are just a way to legitimize compositions. People think that, because a composition was computer-generated, it is somehow more true to the original beethoven than a work completed by a human composer. (This is obviously nonsense; AI has many biases of its own, and purely AI-generated compositions very clearly fail to grasp musical nuances apparent to human composers.) Beyond that, AI-generated compositions tend to attract more media attention, so the computer scientists/composers involved in them tend to overstate the role of AI/machine-learning in generating these compositions. It's practically a cottage industry at this point, and your skepticism is entirely warranted.
Agree, at its mathematical best the most perfect outcome would be "average". This is a simple fact of how this technology works. The miracle of human achievement is its variability and creativity - which is not part of these procedures and WHY we listen to Beethoven in the first place.
It reminds me of a colleague in an introductory AI course. The program she wrote was to bake the Perfect Chocolate Cake. She basically had a fixed list of ingredients and would bake a chocolate cake, submit it to a taste test, and fiddle with proportions for the next round, based on comments she received. She wound up with a cake that tasted like a Duncan Hines cake mix. She eliminated me early from the tastings because I was always the outlier. I preferred a European torte. I don't like American cake mixes.
I would like to express a disagreement. Not all algorithms rely on auto-generated music in Markov, especially with the naïve way that many times referred to. Of course, there are a Great number of improvements that have to be done to model the musical creativity, but It is oversimplified to say that "it is just a Markov model". Long-Short memory, automatic GTTM analysis, and many others can address the problem of local probability dependence.
Politely!
How about now?
Insightful, funny, snarky, and above-all, right on the money, as usual. No AI will ever attempt to create like David Hurwitz.
After listening to the AI "completion" sample, I can only quote the solo bassoon at the end of the second movement of Haydn's Symphony No. 93.
Thank you for restoring reason!
When the austrian composer Gottfried von Einem was asked to finish Bruckner's Ninth, he said no to the project, but used the material for a work he called "Bruckner-Dialog". Luciano Berio did a similar thing with Schubert in "Rendering". I guess, that's a creative way for dealing with fragments, sometimes with more, sometimes with less fortune. But to finish a work, which, in fact, never was really started, and calling it, at least in the subtext of labeling it "10th symphony", "BY Beethoven" is, as you say, hybris. It's one of the greatest chuzpah, I ever heard of.
The mind boggles at what AI could give us. I've always wanted to hear the Délibes Banjo Concerto. I believe there are no surviving sketches, but hey, that's no problem.
As an engineering and computer science major, I find the idea of "finishing" an unfinished work to be completely absurd in general, but especially with AI. I'm curious to know if they even tested their algorithms on other Beethoven works.
The only way to get an idea of accuracy would be to feed their program fragments of a known work of Beethoven's, and then see if what it spits out is close to the real thing.
Forget all the intricacies orchestration, I wonder if this thing could even accurately reconstruct one of his Piano sonatas (which are still enormously complex works).
Someone once pointed out to me that "Logic Omits Us" is an anagram for "musicologist". Coincidence? I think not. Thank you for voicing out some reason.
P.G. Wodehouse (author of the Jeeves books) left us a clear account of his creative process: "I just sit at my typewriter and curse a bit". Beethoven's creative process was probably the same, except that he didn't use a typewriter, and he cursed more than a bit. I doubt AI can use that process to give us Beethoven's 10th, any more than it can give us "Jeeves and the Pandemic Shutdown".
Perhaps a century from now, cyber archeologists will unearth the Ask Jeeves search engine and use it to recreate all the Netscape-designed web pages that populated GeoCities.
we've had Beethoven's 10th for some time - it's called Brahms' First
Oddly enough, that sample of the AI-composed "Beethoven's Tenth" rather reminded me of Brahms.
Good one!
Problem solved.
I read the article you critique a day ago myself, and your deconstruction (or should that be decomposition?) of this more than slightly pretentious and ultimately sterile project is spot on. From musical juvenilia such as his completed-but-surviving-only-in-fragments Oboe Concerto in F Major to the barely begun String Quintet in C Major he was working on at the end of his days, we can only mourn what we have lost of Beethoven's music, and what he might have written during a longer life. Far more importantly, however, we can be infinitely grateful that such a genius ever existed at all, and that he gifted humanity with so many masterpieces that we do, and always will, have and cherish.
On a personal note, I just realized that this is essentially the 25th anniversary of one of my earliest published (in the magazine Analog Science Fiction and Fact, October 1996 issue) science fiction stories, "Symphony in a Minor Key." It is a combination time travel-alternate history tale in which an admiring musicologist from the future changes the past and gives Beethoven seven additional years of healthy life. Naturally, one of the revived composer's creations is the Tenth Symphony, although in this imaginary tale it is an entirely de novo work and not based on the E-flat Major--C Minor sketches. For obvious reasons I could only describe the barest details of what the symphony was like and how it sounded. My novella focuses instead on what impact the Tenth had on the subsequent history of music and the world. (Spoiler alert--to put it mildly, things do not turn out well...)
I remember when the "Jena" Symphony was Beethoven's 10th. I believe it was proved to be by someone named "Witt". There was a recording or two.
The real hubris, as you point out so entertainingly, is the assumption that you could capture Beethoven's creative process at all, even from the entirety of his surviving work. Is that what a creative process is, the sum of parts, something you can put in a box? I don't think you can separate out anyone's creative process from external conditions and circumstances, even something as trivial as what you had for breakfast, or what the weather is at the moment, or reading Goethe at bedtime. I agree that the project is fascinating, but it cannot justify claims to being Beethoven. Sheesh!
It's a question of marketing.
AI Nerds: We taught the AI to take some fragmentary Beethoven sketches and turn them into something that sounds vaguely like Beethoven.
Musical Public: Good for you, I guess.
AI Nerds: We taught the AI to write Beethoven's 10th Symphony!
Musical Public: Wow! Let's hear that!
Karl Holz, violinist of Schuppanzigh quartet and a good friend of Beethoven at his last years, claimed that he played the first movement on the piano. Prior to his death, at March 18, 1827, Beethoven wrote to Ignaz Moscheles that he "is still hoping to thank the London Philharmonic Society for costless help and to dedicate his new symphony which sketches lie on his table, or the new overture, or something else".
From the way the article is written (like explaining what is a fugue and a coda, what does orchestration mean, and using cliched phrases like "taking the baton from Beethoven"), it seems that the 10th symphony was composed to please the non-musicians... like "Ah, you like the ode to joy, here's the 10 symphony composed by AI."
It's hard to think of a more unpredictable composer in the symphonic form. He didn't follow a beaten track, he created a whole new road. As you say, who could possibly predict what came next? And then of course there's that tricky thing called genius. Interesting to hear that AI was getting better at that!
Thanks David for this hugely enjoyable video. Great to hear your lethal, articulate, humorous scorn put to such great use, dismantling this this absurd enterprise. Great stuff.
I don’t understand why they don’t call it something but like “Symphonic Fantasia on sketches of Beethoven.” Clearly if 99% of it wasn’t written by Beethoven it can’t be said to be by him, or to merit a place within his numbered symphonies.
Very interesting point about a Beethoven being an evolutionary composer. Maybe you'll do a video on evolutionary composers vs. non-evolutionary composers.
I hope Currentzis records it.
He'll just rush it.
Oh dear, I read about this. And I heard the excerpt. It sounds like a bunch of Beethovenian themes, lifted from his other works, strung together into a jumble which goes absolutely nowhere. As an AI experiment, it's interesting. As a musical experience, it's pointless, IMO.
David, I don’t have the time to listen to all of your video at the moment, but your title alone made me spit out my morning coffee from a laugh! Thank you, I so needed that laugh.
Nice one Dave - there's a bigger point about AI and creativity but applied here you have nailed it. Much better than bothering with any 10th is spending your time with the 7th or the 4th or the 2nd ............etc
Love your videos David. They’re extremely enlightening and just so much fun.
I was wondering: could you do a video on Beethoven’s Missa Solemnis? I couldn’t find a video on your channel. Hoping you tackle it soon!
Noted
They started shooting "Casablanca" before they had a completed script, and the actors didn't know how the story was going to end. Ingrid Bergman didn't know which of the two leading men she was supposed to be in love with. An AI-generated completion would have had her run off with Bogart, because he was the bigger star, and that was how Hollywood usually decided things. It would have ruined the film overall, but it fit the usual formula, and that's what AI prioritizes.
For sure your funniest commentary yet. I loved it start to finish. How people come up with this nonsense!
Maybe we could get an Eighth and Ninth Symphony of Sibelius and Prokofiev in the process, the 28th piano concerto of Mozart, a sixteenth string quartet of Shostakovich...
Beethoven broke the rules and got away with it, and that is the mark of true genius.
At the present state of AI, I would expect such a project to produce a few passages or perhaps a short movement that sounds "in the style of Beethoven". But the notion that it could start with a few sketches and generate something resembling what Beethoven would have written had he survived long enough to complete Symphony no. 10 is difficult to take seriously.
A test of how close the AI could come would be to give it similarly brief sketches from the composition of one of the completed symphonies and then see whether it produces a piece resembling the real thing (after first deleting the completed version of that symphony from its database to prevent cheating!)
For some reason, this reminded me of the efforts of Rosemary Brown to generate new compositions by dead composers while acting as a spiritual medium. I think that was back in the 1970s. I only remember her name because I knew someone else with the same name and sometimes asked her how the composing was going.
Could AI have created THE NINTH or Quartets 13 & 14? Or the last three piano sonatas if they had all the works written earlier before they had existed?
I wrote this at the start of the video and see you covered these points later.
While I’m at it, how about an AI shot at completing Turandot or the Mozart Requiem and compare with what we have now?
If the experimenters had possesed just an ounce of reason, they would, before embarking on "the 10th", have started out feeding their computer all information pertaining to Beethoven's symphonies 1 through 8, their (alleged) information about Beethoven's composition processes, and plenty of thematical material from symphony no.9 - and then taken a look at, whether their IA was able to come up with anything that was just close to the actual 9th.
I can't imagine that happening - judging from the sound snippet of "the 10th" in the page, that David links to, I'm actually certain, they wouldn't by any standard get even close. It sounds like a hotch pot of bits and pieces from earlier Beethoven ideas, albeit structurally and developmentally awfully realized and in such an amateurish manner that it would have sent good ole Ludwig fuming with rage or brought him to the utmost brink of despair!
@@jensguldalrasmussen6446 Exactly my argument. And that scherzo, as I said, was unspeakably banal. Science and experiment demand controls. Giving AI the sketches of an existing work and seeing what it came up with would prove that the claims made for the Tenth are ludicrous. Though as Dave says, experimentation with AI in itself is just fine--but without such gimmicks.
One of your best, ever, chats. Not one of the "great minds" putting this together, could, with any certainty, predict the next 3 minutes of their own lives, based on the sum total of their own past; but are telling us they could predict what Beethoven would have composed, based on the sum total of his output.
There should be a special place in hell for people trying to create a 10th symphony by Beethoven.
Worse yet, force them to listen to it.
"Virago of Nonsense"! You've just given it a title!
Superb analysis
Leonard Bernstein once said that Beethoven’s greatest strength was his sense of form: everything that happens in a Beethoven symphony happens at exactly the right time, is the only thing that could possibly happen at that point (or something like that). I listened to the excerpt that Dave included a link to and found it to be a complete mess and muddle in terms of balance and form - totally lacking the quality of inevitability that Bernstein noticed and praised in Beethoven’s writing.
I've just come across a new work - we should be grateful it was headed up as 'new' I suppose - Richard Wagner in Venice, composed / realized / imagined by composer Mathew King where he has supposedly knitted together fragments of orchestral works that had been begun by Wagner. Just let them rest!!!
Great video and wow! I never knew people were attempting this sort of thing. This is what I believe in regards to this subject: only Beethoven can write Beethoven... and he's dead...! so there's no more Beethoven to be had. As much as I love him, its as you said... we've got enough of his stuff. I loved how you said "the ninth" with your previous video on "the ninth"!! :-)
When I heard the Cooper 10th, my first thought was that I was actually listening to Ferdinand Ries...
A musicologist named Gerd Prengel constructed movements from Beethoven's sketches that I believe was the most "Beethoven" sounding piece that was not actually written by the composer. (Especially the "Scherzo/Presto".... Although it does have a similar character to the second movement of his Ninth.) He has these compositions posted on UA-cam. I don't think he used AI to construct these movements.
I guess a computer's deaf too. So maybe there's hope
These "symphony 10" recordings are gimmicks, nothing more than that. Shameless, agree!
I agree with all you say here Dave, including the respect for trying things with AI. Also think, as this is a general issue raised by this channel and the other review magazines/ sites, there's the question of how much is enough?! These completions (Levin more recently and others) are particularly noticeable in Mozart. Every anniversary we get more and more little works and fragments completed or not. Didn't we have enough already? I mean there's tons of completed Mozart and Beethoven that if I am honest, I can hardly find the time to be always working my way through and listening to! There are exceptions that prove the rule (Berg's Lulu is definitely one, with of course vastly more basis in what was sketched in short score). But wouldn't it be better is we focused more on writing new music? Or even getting machines to write new music to hear what happens?
Your video prompted me to give a listen to the two movements, they are up on UA-cam. They sounded to me like a grad school homework assignment--here are some Beethoven sketches, write two movements of music in Beethoven's style. The AI's achievement was astounding for AI, fairly mediocre for a human who might have tried the same thing. I thought the movements had a lot of oddly dissociated fragments and strange direct quotations (5th symphony and Pathetique Sonara) that Beethoven would never have done in a million years. And why all the organ music in the AI finale? Is there some historical evidence for this? It sounded just weird to me. I think the movements are an impressive achievement, but only if billed as "An AI Fantasy on Sketches by Beethoven."
We don't need a Beethoven 10th, but I bitterly regret Sibelius never released his 8th Symphony.
100% agreed. Although I was less kind in adjetives with the article’s author.
I once read the complete poems of Yeats. Since he seemed to write in clumps of poems, every so often I'd say to myself, "Oh, I see how he's doing this. It's a matter of certain key words and narrative strategies." Then he'd do, in Monty Python's phrase, Something Completely Different.
The excerpt was interesting, but not particularly Beethovenian. The closest it came was certain harmonic progressions in section A. Even here, however, it sounded like pastiches of other works. It became a game of Name That Beethoven. The B section had the same problem, but the material and argument were hapless, sometimes ridiculously so.
Nevertheless, as David points out, to me it's quite a feat to get the computer to do make even minimal musical sense.
We’re all entitled to her opinion and a number of good points have been made. But I think the effort is still worth it even if what comes out it’s not quite Beethoven. We have to be inspired to do the impossible.
Not this way.
Fot the sake of verification they should do this the other way around. Reduce the Ninth to the most basic themes, more or less to the same level as the supposed "sketches for the 10th", give it to the AI, and see if it comes up with anyhting remotely similiar to the actual Beethoven 9th.
It might be interesting, as a test of the claims for this AI methodology, to submit to this team preliminary sketches of a completed Beethoven symphony, say the "Pastoral", gleaned from whatever musicological libraries have them in their collection. Let them ply their digital magic and then compare the result to the actual 6th. I'd like to keep an open mind, but I suspect that the difference in form and quality would be substantial.
And "two movements of more than 20 minutes"??? Are they sure that someone didn't mischievously slip the dreaded Bruckner virus into the software?
Excellent chat indeed, with which I am in total agreement. I'm sure Beethoven abandoned many more sketches than he ever used (Remember Bernstein's enlightening and entertaining TV lecture on the rejected sketches for the Fifth?). The arrogance of the people involved in this project is beyond belief. At least Deryck Cooke, when, with much much more tangible musical material to go on, he presented his Mahler Tenth projects, had the modesty to call them "performing versions."
As for anyone craving more Beethoven, I admit I sometimes do--but middle-period Beethoven. As valuable as Fidelio is, considering all the time, effort, squabbling, revising and rewriting that went into it, I often think I would trade it for a couple more Beethoven symphonies, concertos or sonatas.
Are you kidding me? The guy who wrote the Intel sustained note is the composer who got that gig? If you want something novel and faithful you'd have a better chance putting it into the hands of heavy metal and acid rock musicians who are steeped in the Bach Rock tradition, starting with the late Frank Zappa.
Listening to the clip, it sounds more like a sketch for the first movement of the 5th symphony.
Feed the AI everything about Beethoven as they did, teach it everything, then feed it ALL the surviving sketches of any Beethoven symphony, but not the symphony itself. Watch it produce NOTHING like the actual symphony that exists, I guarantee it.
I just heard the scherzo sample from the link, it's unspeakably banal.
I've said elsewhere, if people want to feed some Beethoven sketches into a neural net and record whatever it churns out, I have no problem with that and in fact I find it kind of interesting. But my god, don't be so pompously daft as to call it Beethoven's Tenth Symphony!
Exactly.
We are at the 10th already? Still looking forward to your Repertoire videos on the 1st and 2nd :-)
My whole outfit would be made of irredeemable chutzpah
Love your comments.
Even if there were more sketches left about Beethoven's first conception of a tenth symphony, no one, not even an AI, could estimate how Beethoven would have shaped the final work. All his late work, sonatas, string quartets, the 9th, late variations, even smaller works like the Bagatellen, contains formal experiments that make his progress in future projects completely unpredictable. It's just a technocratic stunt, but not a serious scientific musicologal progress.
Indeed. If there was anything predictable about Beethoven, it was his unpredictability.
Goddamn be those PR people. Because it would perfectly interesting to listen to the “Symphony in Beethoven style” by BOThoven 1.0 as a technological experiment. Because that’s the interest: a machine can emulate someone’s style. Emphasis on machine. Because any mediocre student could have composed that faux-Beethoven scherzo.
The Chinese phone manufacturer Huawei once created an A.I. to finishes Schubert’s “Unfinished”. They even hired Alondra de la Parra to conduct its premiere. It made a big splash on tech circles but musically it was ridiculous: the completion sounded nothing like Schubert and absolutely nothing like the preceding movements. In fact, it’s so happy and cheesy that it would be perfect as the soundtrack for some “child-dog movie”.
This is a good start. But I want to know if the computer can give us some more Beatles albums?
Here's a suggestion which, if followed up, would keep the ASFI experts hurtling round the world for years to come. Feed a computer with every orchestral work by Beethoven from the second symphony onwards. Then get it to reconstruct the first symphony. Then compare and contrast. The possibilities, if unimpeded by common sense, are endless, and I look forward keenly to Hildegard of Bingen''s saxophone quartet, and Josquin's Missa "Sophisticated Lady". ASFI stands for "Artificial Substitute for Intelligence".
Will ASFI recreate P.D.Q. Bach's works?
I would not want to see the sacred memory of PRQ traduced by a bunch of computer nerds. We must draw the line somewhere.
To participate in what sounds like a fabulous junket, I would not hesitate to help complete not only Beethoven's 10th symphony, but also his 11th and 12th!
Like I said...
AI should attack Bruckner! Give the world another Bruckner symphony...on second thought.
Teaching the AI about Beethoven's musical processes isn't enough. To get convincing fake Beethoven you have to teach the AI to be German, deaf and cantankerous, live in Vienna in the 1820s, simultaneously admire and strive to surpass Haydn and Mozart, pursue unsuccessful love affairs with a string of women, contemplate suicide, sue for custody of it's nephew (first teaching it what a 'nephew' is), etc., etc., etc.
And never bathe.
@@DavesClassicalGuide Interestingly, according to his secretary Anton Schindler, Beethoven washed his hands quite frequently, but he made no mention whatever about the master cleansing the rest of his body. Must have gotten pretty ripe, especially in warm weather. Yet another odd variable for the AI to ponder.
This is so silly. Beethoven left vast swathes of music. The quantity and depth of what's already there are enough for a couple lifetimes. I doubt the people on this project have fully contained the scope of his output either! Seriously nothing more is needed lol
The real problem is: How can you make a computer DEAF? Second: I am waiting for the Portugese Suites from J S Bach.
What's next the 'realisation' of Symphony no.0? As there are sketches for a C major symphony that Beethoven wrote before working on his first symphony, these are not related to the first symphony.
So basically this piece of music should really be called 'Symphonic Fantasia on Fragmentary Sketches of Beethoven'.
What I don't get is why the AI people took a detour to go on bended knee in front of the great man theory of holy german art; Deep Blue didn't become famous for accurately mimicking Kasparov, it became famous for BEATING Kasparov.
This entire project is nothing but a cheap publicity stunt for Deutsche Telekom, the former state-run corporation that single-handedly made us Germans have one of the worst internet infrastructures in the entire Western hemisphere.
I like the Barry Cooper movement sort of. I don't really hear it as Beethoven's 10th, but (considering the form) as a quite original and enjoyable standalone. The 19th century didn't produce something quite like it in its formal concept (although a vast number of much better pieces of course), all the more reason why I find it weirdly refreshing. Probably something you could call a guilty pleasure.
The sample I heard is actually reminiscent of Mozart's musical joke only much, much more hilarious (simply because the music has no sense to it whatsoever).
Think i'll Stick with his one through...THE NINTH!...Symphs.
even if they are just seeing what ai can do, do these people realise the sickening affront to art and human expression these recreations are? i find it amusing
I suppose whenever you mention completions of "The Tenth" you should show up in a Speedo rather than a tie. I think we'd rather forego the results in both cases. lol.
I heard it and I loved it but I do agree it’s not proof that it was gonna be a symphony but either way it’s Beethoven so im gonna take it
Is it?
When you actually listen to the little excerpt, you will notice that it is just a combination of symphonies 5 and 9. Did the authors of this project not listen to their own creation? Pardon me, Beethoven's creation. Do they consider that Beethoven created new things and evolved without repeating himself every second or so! David's talk was entertaining but the music itself was hilarious.
I perfer this AI version of Beethovens "10th symphony" before Barry Coopers.
You can bet that all of these artificially intelligent musicologists will be living it up with the foundation paying the bill. Nothing that is musically artful will result, though. It will be artificial, but not intelligent.
I've just listened to the fragment in the link. Sounds to me more like bottom-drawer Schumann than Beethoven.
Beethoven is sleeping well... I wouldn't bother him
I'm one of those old curmudgeons who thinks unfinished works should be left alone. Finishing an unfinished (or un-begun) symphony might be a fascinating AI experiment, but I think humanity has more pressing matters for which to utilize that technology.
Speaking of curmudgeons, imagine an AI program that was designed to write "new" columns, books, and texts by Oscar Wilde, James Thurber, Mark Twain, H.L. Mencken, and Sinclair Lewis.
Why, that sounds like the plot of a Roald Dahl short story.
@@robertjones447 A Dahl story, or maybe an H.P. Lovecraft wierdie.
Isn't Brahms' 1st known as Beethoven's 10th?
Already been said, but that's not the point, is it?
Does a Beethoven sketch written by Beethoven sound like Beethoven? Wellll....full stop. If the sketch came from the mind of Beethoven it is indeed Beethoven. Having said that, a cluster of sketches, even if Ludwig intended to use every note does not make it a symphony.
AI is killing the arts business, I put it that simple, computers are not human and humans are not computers, can a computer feel emotion for MUSIC? I admit that it is more fun and even tender to see robots conducting orchestras as in the case of Yumi or Asimo, but how do you teach a robot to appreciate Music? How do you force a computer to think or feel like humans? For a reason it is called "Artificial Intelligence".
Let us remember that Beethoven had a long time wanting to adapt Schiller's Ode to Joy (Schubert also made a lieder on the text), And if for something the last quartets have more than 4 movements and the 9th is extensive, it is because I think that Beethoven was testing all his compositional power before leaving something unfinished at the time of death, he already wanted the Ode to Joy to be as a separate chorus, that is why he had the desire to include this chrous and extend Schiller's text in a heavenly way before leaving the world, because he knew that it was going to be his last symphony, who knows how the 10th would have been and towards what directions Western Music would have taken, who knows if it would had a number of bars close to a Mahler symphony, is the same case of Schubert's unfinished symphony, maybe Schubert didn't want it to be performed in public, but I think the fact that it's incomplete is what makes it interesting, In any case, Schubert's style would not have approached so much that of Bruckner (and that both had Simon Sechter as a teacher of counterpoint), and now the AI wants to complete Schubert's Unfinished, Bruckner's 9th or Mahler's 10th and maybe (although I don't want it to happen) Sibelius's 8th.
As if human talent is no longer enough for the Arts or entertainment industries that great works are no longer needed, another demonstration that human beings do not care about legacy and preservation, for lack of instruction and respect for Culture.
A new twofer! Schuberts unfinished and Beethovens unbegon. Next sibelius 8th.
Paul G.
But seriously, folks, why do we do this with composers... and authors (Tolkien comes to mind)? Here is my humble rumination written elsewhere on the idea of musical completions, in this case the work being Mozart's Requiem: "By keeping alive the puzzle of Mozart’s deathbed creation, we are in a sense keeping Mozart himself alive, denying to ourselves that his magnificent opus is complete. We are in essence keeping guard over Mozart in a perpetual wake, refusing to bury 'the miracle that God allowed to be born in Salzburg' and allowed to die too young in Vienna."
Only two reactions possible to the horrible daft enterprise - and I appologize beforehand to the poor cleaning lady (no matter of which sex): 🤮 and 😭 (unfortunately I couldn't find one emoji, that combined the two of them!).
Why bother with artificial intelligence, when you do not posses the genuine article?
The whole business reminds one of a sick joke - and, indeed, I had to pinch my arm and make sure I hadn't been somehow teleported back in time to the first day of the month of April.
The stilted, pompous baloney, that was read from the commentary, immediately brought to mind the fairy tale of my famous compatriote:
"The emperor marched in the procession under the beautiful canopy, and all who saw him in the street or out of the windows exclaimed: "Indeed, the emperor's new clothes are incomparable! What a long train he has! How well they suit him!" Nobody wanted to let others know, that they saw nothing, as they then might have been deemed unfit for office or too stupid. Never before had the emperor's clothes been so profusely admired.
"But he has nothing on at all" a small boy suddenly exclaimed.
Oh, thou ever young curmudgeon, lift up thy voice, reveal their empty posturing, expose them, bare-assed, for all to see! And we will keep on listening!
I'm in agreement with all who here think this is simply just wrong. 1st this is simple...These are men trying to make a name for themselves and this is entirely self motivated. Beethoven or any composer for that matter has the right to keep their music strictly theirs and theirs alone. They share and we listen period. I think this is ridiculous. the greater point is not Beethoven or this music. It's the status hungry that is the real point. No matter what they do it will be wrong. It is and will never be Beethoven's from his head period. arrogance and glorifying themselves. ALso, in closing how utterly disrespectful to the man's heart. Aside from his Genius and all he went through to write these pieces. As a composer myself. I believe they call these pieces because they hold such precious pieces of ourselves that we share with others and to think not only that YOU are capable of taking that and also giving it to a computer? Total rubbish! let them pretend to be something all they want. we can just shake our heads and quietly walk away ...they will have to face the man some day .....so let them live with that alone in the dark when no one is looking.....but him :)
hope this was ok to say....but its offensive to me.
heard a one minute clip of this, sounded like a really rubbish version of beethovens 5th 3rd movement, no doubt that is where it came from
The Hat of Horrible Hubris
In my view, the only way the musical sample at the end of the article might merit any comment is by assuming it was meant as a parody on some of Beethoven's music. But it fails even under that assumption, because it's not funny or witty at all. It's just embarrassing. My message to the distinguished (?) team of experts: Go get a life!
LvB would have never ever used THE EXACT rhythmic motif as in previous symphonies, let alone the augmented one in 3rd mvt of the 5th Sym. (augmentation of the motiv from the 1st mvt). This attempt has traces of Mr. Levin's grandiose narcisism- it is obvious that he chose the computated options ala- it could have been this, it could have been that. The attempt uses motifs from various B' works in a 2nd grade fashion. This is precisy how a composer's mind DOES NOT work. He/she invents new on the basis of previous experiences-just look at B's 32 piano sonatas.
No serious composer would have ventured into this. It is not good, it is way less than a 10th would have been. Go play with some 2nd grade sketches from elsewhere, omg
Having listened to the excerpt, I'm already bored. The music simply doesn't go anywhere, it doesn't progress harmonically or thematically. It's just note-spinning. An insult to Beethoven, if you ask me, but I don't suppose Beethoven would have minded. Fortunately, he wouldn't have been able to hear it anyway.....
Beethoven's 10th would have looked forward toward unexplored vistas as did the late string quartets and late piano sonatas. This AI version is stuck producing mediocre versions of much earlier Beethoven. Garbage, and should be illegal.
I am in the minority who think Beethoven should've stopped after composing his 7th Symphony. His Ninth fell short of the perfection of the 7th, and should have been replaced his 6th, an entirely superfluous composition. I'm not being serious here, except I'm completely serious. Pick one.