Eastwest partners with ACE Studio

[Linos]

Eastwest Sounds announced that they partner with ACE Studio:

https://www.kvraudio.com/focus/ace-stud ... tion-65954

Not long ago, Herb Tucmandl, CEO of Vienna Symphonic Libraries, announced that VSL are researching and working with AI for future libraries.



There is a lot of movement in the world of orchestral samples. After years where innovation seems to have stalled, major changes are afoot.

[Duncan Krummel]

I can appreciate that ACE Studio has taken an approach that stays away from fully generating entire pieces, but to be honest it feels like it’s just passing the puck on from automating the writing to automating the musical expression and interpretation. I don’t want a human-less interpretation of music, even if this finally improves EW’s inconsistency with things like legato transitions. Would love to be wrong here.

[Luke]

I have the same concerns as Duncan.. My attitude for the moment is one of "wait and see." With what we do with mockups, one of the few bastions left in stitching together samples, is the personal musical interpretation of the phrasing and expression. Modeling gives great freedom in that arena. But perhaps there could be a way to work with this technology and integrate our vision of these components in its renditions. That is my hope.

[Guy Rowland]

I can appreciate that ACE Studio has taken an approach that stays away from fully generating entire pieces, but to be honest it feels like it’s just passing the puck on from automating the writing to automating the musical expression and interpretation. I don’t want a human-less interpretation of music, even if this finally improves EW’s inconsistency with things like legato transitions. Would love to be wrong here.

Can open, worms everywhere...

I'm going to struggle to feel nostalgic for lumpy legato or dynamic crossfades. It's always been Frankenstein's monster - the game has always been to make the least monsterish monster. It's not really a hill I fancy dying on.

There is human skill here of course in making dead samples sound as least bad as possible. There is too in, say, the dying art of rotoscoping, but however much I admire the geniuses who can pull that off without ending up in an asylum I don't think I'll mourn the loss of humanity.

I don't really know what these guys have in mind. But if I can command realistic transitions in a way I haven't been able to before, without the burden of interminable engineering pain, I'll live with the bittersweet knowledge that the good old days ain't coming back.

Doubtless the trickery will go far beyond this. I will probably be able to whistle a tune into my phone, press a button and it will sound like Alan Silvestri. But what am I saying? I can do that right now if I want.

I think a dividing line between good and bad AI is juusst beginning to emerge. Vague though it is, if a tool under the command of a creative person helps him or her achieve a better result than conventional techniques, then hello - this could be Good AI. If it can be autonomous and replace the creative process (or claim to), well, that might very well be Bad AI.

Stem splitters? Good AI. Dialogue isolate? Good AI. Sumo? Bad AI.

[Lawrence]

I LOST AN ENTIRE CUBASE FILE!!

I have a songwriter/client who hires me to take his very basic piano/vocal .mp3 (he is self-admittedly NO singer and can barely play piano), learn the tune, orchestrate it, sing and mix it. He asked me to replace a vocal with a new lyric on a song he had hired me to work on two years ago, ...and I COULD NOT find the Cubase file. He sent me an .mp3, and it was clearly me singing and playing.

Sometimes I find Mac search to be oddly unreliable. It's probably user error but regardless, that file was not showing up though I went through my drives forensically. I was really panicked.

Thankfully, the new lyric scanned just like the older one. The solution was simple-AI. Strip the vocal off the file and re-sing. It took me hours to come to that simple solution, because I'm well over one hundred years old and used to doing things manually.

My point (as Guy also opined) is that among the Bad Consequences there are going to be Good and Helpful Things to be gotten from our new AI overlords, and there will be picking and choosing.

I'm not sure about the Ace Studio thing. We'll see.

[Duncan Krummel]

Time will tell, Guy. I do hear you, though. I’m no fan of bumpy transitions, but I’m feeling exhausted by sampling in general these days.

I don’t think I’m ever going to be able to appreciate art separated from the humanity behind it. Give me a human interpretation of garbage before you give me a synthetic interpretation of a masterpiece.

Anyways, I am still waiting to see what EW + ACE brings to the table. I won’t write it off outright, but it doesn’t excite or inspire me. So, I guess in this regard, I fully sympathize with your optimism, Guy and Larry. I don’t share it per se, but I’d also like to be wrong.

[Ashermusic]

I am pretty sure, much where Duncan is, but if I were 57 instead of 77 I might feel differently.

[Linos]

For me, there is a philosophical question. Let's assume that AI will reach a point where it can produce an audio file from a score. And that the performance sounds real. A result that would be much better than what mockup artists can achieve today with far less tinkering. Would this be a good thing?

For one thing, I enjoy creating mock-ups. Admittedly, there are tedious tasks involveld, such as programming around bumpy legato and other limitations of the samples. Nevertheless, I think the resulting audio will be richer if a human has worked on it. Maybe not even perceptibly. But when humans make musical decisions, they imbue the mockup with their specific abilities and sensitivities. Compare this to a similar audio file created purely by AI with little human input, apart from the material it was trained on. Knowing the background stories, which one would you prefer to listen to?

Of course, you could argue that it is not you in the performance if session musicians record the piece in a studio. But humans are still creating the music and making decisions while doing so. I believe that the story behind the creation of a piece of music enriches it beyond what AI can achieve.

If AI were to replace gigging musicians, I would see that as a sad development. However, if AI can be used to improve the quality of mock-ups and we are still responsible for making all the non-obvious decisions, it could save a lot of time and be helpful.

As of now, I have no idea what direction we are heading in. Perhaps it is both directions, depending on the company and tool.

[Guy Rowland]

Imagine a composer has written a full score with detailed dynamics. She then picks up her baton before the Royal Philharmonic, and they're off.

She stops at bar 34 cos they weren't interpreting the score correctly and gives her notes. They're shit-hot otherwise. I mean, they are the Royal Philharmonic.

Then she wakes up from her wonderful dream, penniless, living in a scummy bedsit. She switches on her aging computer and starts fighting with Dorico and Cubase for the third straight day. It has worn her down and, eventually, her will to live surpasses her will to make it any less shit.

Then she wakes up from her nightmare in the year 2029, still penniless and still living in her scummy bedsit. She imports the score into Cubase 19 which has Score2Audio. It sounds pretty good until bar 34 when it didn't interpret her score correctly. She adds a couple of prompts and all is good.

In both dreams and the future, she is the same person - a composer, not a programmer.

[Duncan Krummel]

Not to be overly negative or curmudgeonly, but I do have this growing feeling that things worth pursuing in life shouldn’t necessarily be easy. I don’t mean that it should be difficult to access, but that attaining a high level of skill should require work. I mean this both from a philosophical point of view and a psychological one. It sounds incredibly elitist to say “music should be hard,” but I do genuinely feel this way. Great music takes a lot of internal reflection and dedication to create: it requires discipline, motivation, deep thinking. It’s something I firmly believe everyone is capable of doing and that the barrier to entry should be as low as possible, but the actual doing of the work itself should be rewarding, and that - to me - requires effort. An AI technology that perfectly realizes a score is simply a machine interpreting instructions.

Despite what peddlers of third-rate sociology will tell you, people are not machines. We bring our histories and opinions into everything we do. That’s not a feature of art, that IS art. SynthV, as I wrote in the other section, is a big step forward for me to make mockups, and I fully recognize this is based on machine learning. The mockups are great for purposes of getting actual humans interested, but the output is lifeless and meaningless to me. I’m not a master of anything, but as I enter my 30s, I’ve finally fallen in love with the WORK of practicing, studying, researching, and developing a deeper understanding of art. It saddens me that there are people (not here, mind you, but I see this sentiment on other platforms) so uninterested in the process, only the product.

[Linos]

In both dreams and the future, she is the same person - a composer, not a programmer.

That is true. But will the Royal Philharmonic continue to exist in the second dream? If AI can produce a performance that surpasses the abilities of the finest musicians, will we still attend concerts featuring human performers? Is there something we value in the human pursuit of perfection, even though it is always doomed to fail? Or will we choose easy perfection generated by writing three prompts? To me, the struggle to improve and express oneself is an important part of what makes art valuable.

The same goes for the act of composing in the second dream. If AI can create a composition with little to no human input, would the composer still be considered a composer? Would we still listen to their imperfect composition? Or would we listen to flawless perfection generated by AI?

I fully agree with Duncan. I firmly believe that there is value in doing something that is difficult. Making it easy diminishes the task's value. Perhaps it's comparable to climbing a mountain versus taking a funicular railway to the summit. In the end, both people will be at the top enjoying the same view. However, I believe that the person who climbed the mountain is richer for the struggle and that their accomplishment should be valued more highly.

This does not mean that the funicular should not be built at all. If it can give people who would otherwise be unable to reach the mountain the opportunity to enjoy the view, that's a good thing. However, if everyone took the funicular and the walking paths fell into disrepair, it would take a valuable experience from us.

[Guy Rowland]

I think (with the greatest of respect) you're both missing the point.

The point is that all the skill of my fictional composer - I'm going to call her Melody - is in COMPOSING. We kinda jumped the whole part of "the fully finished score complete with detailed dynamics", didn't we? That's the human part, no different for Melody than for Beethoven (except Melody's score is all on a screen rather than parchment). And if Melody were here with us now, she'd tell you just how much blood sweat and tears went into that score (not to say talent). Don't call her workshy, she'll kick off.

Beethoven never played a note of his scores. We've gotten used to this strange world with VIs where composers are now somehow also expected to be programmers all in service of producing something doesn't sound nearly as good as a bunch of talented people in a room 200 years ago. Why is that now essential?

Luke, you mentioned "if AI can create a composition with little or no human input" - well, according to my thoughts earlier, that would be Bad AI. Melody is a brilliant and talented composer, the thought of AI writing notes for her - or orchestrating even - is of zero interest to her. She is using Good AI to bring her score to (artificial) life, and she's far happier in 2029 than 2026.

But Cubase 19 only gets her so far - no-one wants to sit in an auditorium and listen to her press play. Live music is still where musicians are essential. That's still her dream.

[Luke]

Perhaps you meant Linos, Guy?

Regardless, in your hypothetical scenario, it is great for Melody to have the possibility of having her score brought to life by technology. And point taken, that we as composers today are stuck in a "VI" pattern that requires us to interpret our own composition.

As to the player issue, and the Philharmonic by extension, my thoughts would be that anyone who's ever given a piece of their music to a ridiculously talented musician to play, will know that their interpretation brings it to life. That is to say, you can mark down every nuance of the imagined music into the score and still be surprised by what someone else plays. The page is an imprecise, subjective (and often dictated by others' taste and personal history) affair. This is why Toscanini's Beethoven sounds so vastly different from Karajan. That particular skill set that every musician earns is absent from AI. I think that should bear more weight than we are assigning it right now and goes into the point Duncan and Linos are making, which to me is very valid as well. Not because of elitism, but because it must mean something, to be able to imbue yourself into a piece of music, art.

[Duncan Krummel]

I think (with the greatest of respect) you're both missing the point.

The point is that all the skill of my fictional composer - I'm going to call her Melody - is in COMPOSING.

Actually, this is where I think we disagree, Guy. Composition isn't one skill, it's an elaborate and diverse set of skills. What specific skills are in vogue and what specific skills are outdated are probably something the global collective will never agree on, but outsourcing a portion of these skills to a computer is bleak to me. Very bleak. I want to BE a great composer, not just sound like one.

Beethoven never played a note of his scores. We've gotten used to this strange world with VIs where composers are now somehow also expected to be programmers all in service of producing something doesn't sound nearly as good as a bunch of talented people in a room 200 years ago. Why is that now essential?

I don't think it's essential. I think this way also leads to some kind of "half-art." I despise sample-based music with the kind of fervor only someone fully entrenched in it can. It's barely able to do what it does best, and the music I used to want to write, and now do out of defiance, isn't the kind of music samples OR machine learning will ever figure out how to do justice to. It's just not what music is. In my opinion, of course.

Regardless, in your hypothetical scenario, it is great for Melody to have the possibility of having her score brought to life by technology.

I agree with this, and with Guy's sentiment. Truly. It's hard enough for me to find musicians of any ilk willing to take on new music, but that's largely geography. For me, being able to realize some kind of approximation of my music is something I deeply appreciate, but I can barely listen to it. It sounds sampled. All of sampling sounds sampled, and AI, I think, has been the impetus for me to lose my taste for the whole thing.

Anyways, I really appreciate everyone's insights. I'm too often given no room to work these problems out with other great minds to challenge me on it, so for that I am beyond grateful.

[Lawrence]

Over the course of a long career in music production and performance, I’ve been guilty of putting plenty of musicians out of work with orchestral samples, drum machines, sequencers etc. Early on I knew there’d be a day when they came for me, and indeed that happened through licensing, DJ’s instead of bands etc.

I have any number of personal projects going on now and I still retain a handful of paying clients, mostly older than I am and less plugged into tech. The skills I gathered over the years allow me to help make their dreams come true, which makes me feel good.

I feel blessed to be at a point in my life and career to (mostly) leave the “what’s coming next??” worries to younger folks, as whatever it will be probably won’t help me much nor threaten my livelihood or my ethos.

[Guy Rowland]

Perhaps you meant Linos, Guy?

Regardless, in your hypothetical scenario, it is great for Melody to have the possibility of having her score brought to life by technology. And point taken, that we as composers today are stuck in a "VI" pattern that requires us to interpret our own composition.

As to the player issue, and the Philharmonic by extension, my thoughts would be that anyone who's ever given a piece of their music to a ridiculously talented musician to play, will know that their interpretation brings it to life. That is to say, you can mark down every nuance of the imagined music into the score and still be surprised by what someone else plays. The page is an imprecise, subjective (and often dictated by others' taste and personal history) affair. This is why Toscanini's Beethoven sounds so vastly different from Karajan. That particular skill set that every musician earns is absent from AI. I think that should bear more weight than we are assigning it right now and goes into the point Duncan and Linos are making, which to me is very valid as well. Not because of elitism, but because it must mean something, to be able to imbue yourself into a piece of music, art.

Sorry Luke (and Linos) - yes I did mean Linos. (Which I just typed as Lions, so the senility is strong with me today).

Well of course I will counter that a facsimile of that can be coded. Prompted. Indeed, that could be one alluring aspect of AI that it might have the nuance of a particular orchestra / conductor in a way that is absolutely impossible for samples no matter how much time a skilled person can put into them.

Samples.

I just have a feeling that 50 years from now conventional sample libraries as we know them will be long gone, a transitionary phase of technology. Sampling itself will live on to be used in creative ways - I mean specifically breaking down the performance of acoustic instruments into tiny chunks that can be reconstituted in some form to mimic the real thing.

Samples have given us this miracle of sounding a bit like the real thing. In many contexts most ordinary folk wouldn't know the difference. And it's alluring for us, isn't it? Under our fingers and mouse clicks we can do things inconceivable 50 years ago. Just one person, time, skill and talent can do that. So I can see why it will be mourned, like all technological eras eventually get mourned.

They just re-released the Commodore 64 you know, when life and tech was much better. Got a 5 star review in The Guardian, that.

[Guy Rowland]

Actually, this is where I think we disagree, Guy. Composition isn't one skill, it's an elaborate and diverse set of skills. What specific skills are in vogue and what specific skills are outdated are probably something the global collective will never agree on, but outsourcing a portion of these skills to a computer is bleak to me. Very bleak. I want to BE a great composer, not just sound like one.

But composition - and arrangement - is on the stave, no? Beethoven played by enthusiastic 8 year olds will likely not fully do justice to his genius. The best orchestras will showcase his brilliance to the fullest extent. But they're both working from the same sheet music.

The idea that a composer must be more than this, that being a composer somehow means executing that score themselves, is terribly recent and surely wrong. Daryl here might have something to say about that.

Melody knows that Cubase 19 is just a trick, albeit a very good one. And she would give her right arm to be standing in front of the Royal Philharmonic with her music on their stands. But purely as a composer - as opposed to composer/programmer - her work would be the same.

[Duncan Krummel]

But composition - and arrangement - is on the stave, no? Beethoven played by enthusiastic 8 year olds will likely not fully do justice to his genius. The best orchestras will showcase his brilliance to the fullest extent. But they're both working from the same sheet music.

I don’t think I agree, no. Composition and arrangement are in the mind, in the hands, and on the stave. And among these are countless individual skills. Inspiration, theme, development, register and voicing, accompaniment, depth and layering… so many countless things to practice and improve. Just calling it composition or arrangement belies the innumerable different factors and skills that go into it. Specifically regarding Beethoven, he was a skilled singer and pianist. He absolutely could perform his own music. An 8 year old most likely lacks the physicality, technique, and wisdoms to move me the same way an experienced musician would, but it would move me far, far more than a computer would (which would be not at all).

Besides, are any of us an orchestra? No. By its very nature, an ensemble is multiple people, each with their own background and intuition. That matters much more to me than the notes on the page.

The idea that a composer must be more than this, that being a composer somehow means executing that score themselves, is terribly recent and surely wrong. Daryl here might have something to say about that.

Historically, I think that was much more the case than today. I’m not a purist, or conservative musician by any means, but I do think being a competent musician requires more than simply arranging sounds. I think a great proof of this is how simply learning an instrument makes everyone I’ve ever had the opportunity to observe this with a better composer. Myself included. The hard work makes us better musicians, and better people.

I would be curious, to all, not just Guy, to see what composing looks and feels like to you as individuals. I think we are many times blinded by our own personal experiences and definitions of an activity. I could very well come out the other side with a changed mind, but for me the work of a composer is so, so much more than putting notes on a page and letting other people (or indeed, machines) do the work of interpretation.

[Guy Rowland]

I don’t think I agree, no. Composition and arrangement are in the mind, in the hands, and on the stave. And among these are countless individual skills. Inspiration, theme, development, register and voicing, accompaniment, depth and layering… so many countless things to practice and improve. Just calling it composition or arrangement belies the innumerable different factors and skills that go into it.

But but but but - for Melody, all of those examples are seen on the stave. That's my entire point.

Specifically regarding Beethoven, he was a skilled singer and pianist. He absolutely could perform his own music. An 8 year old most likely lacks the physicality, technique, and wisdoms to move me the same way an experienced musician would, but it would move me far, far more than a computer would (which would be not at all).

Besides, are any of us an orchestra? No. By its very nature, an ensemble is multiple people, each with their own background and intuition. That matters much more to me than the notes on the page.

But but but but - you are not comparing like for like here. We have to compare AI with dead samples, not 8 year olds.

This is the fundamental point I feel keeps getting lost. We have seemingly made our peace with being Frankensteins, trying to engineer humanity from tiny inanimate bits and pieces. We feel that process is "human" but there isn't a scintilla of genuine musical performance in it - it's all just a trick. (Again, ask Daryl).

That's not to say that Frankenstein's monster is not a phenomenal achievement in its own right. Frankenstein can build all manner of weird and wonderful things that entirely justify their right to exist. But he can never build a human or create a soul.

Oh look, we're back at AI again, of which the exact same principles apply. Don't believe Chat GPT is your romantic life partner, folks.

Historically, I think that was much more the case than today. I’m not a purist, or conservative musician by any means, but I do think being a competent musician requires more than simply arranging sounds. I think a great proof of this is how simply learning an instrument makes everyone I’ve ever had the opportunity to observe this with a better composer. Myself included. The hard work makes us better musicians, and better people.

I would be curious, to all, not just Guy, to see what composing looks and feels like to you as individuals. I think we are many times blinded by our own personal experiences and definitions of an activity. I could very well come out the other side with a changed mind, but for me the work of a composer is so, so much more than putting notes on a page and letting other people (or indeed, machines) do the work of interpretation.

I don't have a right to call myself a proper composer in the sense we're talking about here. I'm no Melody (she's a phenomenal pianist btw). I can barely read music, I never think in terms of a stave, my keyboard skills are lamentable. I am that dreaded and oft-ridiculed phrase a "modern composer" who can use a ton of computing and engineering skills via a DAW and VIs to create things that hopefully sound okay. I do little orchestral these days.

Perhaps Piet here represents all a "modern composer" can be. He has spoken movingly about how a DAW is his primary tool, His ability to take dead bits and pieces from here there and everywhere and make phenomenal music from it is truly something to behold. I shouldn't attempt to speak for him, but I do remember him saying on multiple occasions actually trying to simulate an orchestra holds almost appeal to him these days, the task itself is inherently frustrating and foolish. I feel this is the essence of what we are talking about here.

This discussion has arisen because of East West and their orchestral libraries, the potential for one labour-intensive clumsy trick to be replaced by another near-effortless one. I haven't heard anything yet that says that for someone who has worked for decades on their own playing, composition and arrangement skills when writing for orchestra, this will represent some kind of reduction of their own musical humanity or identity. Melody is not forsaking the Philharmonic or even the 8 year olds - she has simply abandoned being Frankenstein, having never really wanted to be him in the first place.

EDIT - and to post-script - Melody is under no illusions that the finished result represents human performance. It's still a trick, just a more elegant one that involved outsourcing those technical skills that are not composition or arrangement.

[Linos]

Guy, I do get you point. Our opinions differ, I think, on the point how we value the creation of mockups. To me, it's a craft that has value in itself. It's expressing musicality through samples. The very best mockup artists do perform with samples in a musical. They play in musical phrases. You seem to view the creation of a mockup as mere programming, as a necessary means to an end without any intrinsic value. Is that correct?

[Guy Rowland]

Guy, I do get you point. Our opinions differ, I think, on the point how we value the creation of mockups. To me, it's a craft that has value in itself. It's expressing musicality through samples. The very best mockup artists do perform with samples in a musical. They play in musical phrases. You seem to view the creation of a mockup as mere programming, as a necessary means to an end without any intrinsic value. Is that correct?

Oh no, it's absolutely as you describe it.

I guess what is happening is we're trying to get close to what is in our heads. And great libraries can surprise and inspire us - this happens to me all the time that these clumsy tools actually become part of the composition process itself. But that could be equally true of AI. If I'm playing a gorgeous V1 patch using AI vs samples it's not a whole lot different to using LASS except it has the potential to sound better.

That's how I would use AI cos I don't write notation. Melody would skip all that, she's a proper composer! Really what I'm arguing is that whether or not that tool is AI or dead samples, used in this way it's a detail. I don't really know cos I've never used it, but didn't Sibelius have a crap orchestra built in? Plenty of proper composers used that just for rough demos and would bitch about it. AI would just do it a lot better. The end result is no more a true human performance, but it does sound a lot more pleasing.

[Duncan Krummel]

But but but but - for Melody, all of those examples are seen on the stave. That's my entire point.

Right, but I disagree with the basis behind that point. Melody might see that as being all on the stave, but then I would say she is mistaken. That is a reductionists view on what composition and arrangement are.

Maybe I’m still misunderstanding, and apologies if I am, but the written score doesn’t capture everything inherent in a musical performance, nor should it. To take the role interpretation away from the interpreting musician is folly.

But but but but - you are not comparing like for like here. We have to compare AI with dead samples, not 8 year olds.

Perhaps I misunderstood, but you brought up the example of Beethoven and the 8 year old in order to illustrate that the same sheet music can be interpreted differently, right? Then that’s exactly my point. The ‘facsimile that can be programmed’ is a snapshot of a single interpretation. AI cannot think, because it’s not actually intelligent or conscious, and any argument that it will be in the future isn’t founded on anything we actually know to be true. It’s a hypothetical.

So AI anything is incapable of novel interpretation. Whether that’s better or worse than samples is also not an inherent truth. Right now samples do a BETTER job at interpretation, IMO, because a human has to actually attempt to craft one into the performance. Maybe these EW/ACE products will just be - for all intents and purposes - sample libraries like we all know and maybe love, but with improved coding and transitions provided by machine learning. But we don’t know, and I am highly negative on the whole thing. I have yet to see a successful integration of AI that doesn’t sit like a heavy stone in my stomach.

Anyways, not to harp too much on it. I just see it all as a spectrum of human involvement. At one end, a real person writes with their own ideas and inspiration, and a real person interpretes it. At the other end, a prompt generates a facsimile of human creativity. In the middle, someone is removed, and someone remains, but there is still a real person there. I’d prefer the first, I do not consider the second art at all, and I am increasingly uninterested in the third. Art requires a person, and art requires effort. Otherwise, it is art in appearance only to me.

[GR Baumann]

Long ago somewhere on TSB, I expressed my opinion concerning AI as a musical technology being the next step forward, leaving awkward crutches such as key-switches finally behind.

Perhaps that time has arrived. Back then I think I envisioned a method where LLM would be continuously fed with all available musical data, and from there AI to choose & suggest parameters such as articulation and/or dynamics while improvising such on a masterkeyboard.

A complex undertaking, and I do not expect the first iterations of such or similar technology to blow my mind, but ultimately it is a direction I personally welcome, as long as the final decision remains with me, of course.

Somehow I was expecting that to be developed in conjunction with a DAW, now it looks like East West is first to embrace the opportunity.

[Guy Rowland]

Melody might see that as being all on the stave, but then I would say she is mistaken. That is a reductionists view on what composition and arrangement are.

I'm not sure how you can argue this historically. Beethoven has been dead since 1827. If an orchestra plays it from that day onwards the stave is the only option open to them. It was the language of music that Beethoven used to compose and arrange. One conductor and one orchestra can add their own interpretation on that of course, but from the original composer's perspective their notation was their complete intent.

The idea that composers can be more than that is a modern one. And of course they can and most are (if I am a composer at all, this is what I am). But, again, we are talking about the specific job of the orchestral composer. Being able to mock up is a modern luxury that an orchestral composer is not bound to use at all (yet again, Daryl).

Perhaps I misunderstood, but you brought up the example of Beethoven and the 8 year old in order to illustrate that the same sheet music can be interpreted differently, right? Then that’s exactly my point. The ‘facsimile that can be programmed’ is a snapshot of a single interpretation. AI cannot think, because it’s not actually intelligent or conscious, and any argument that it will be in the future isn’t founded on anything we actually know to be true. It’s a hypothetical.

So AI anything is incapable of novel interpretation. Whether that’s better or worse than samples is also not an inherent truth. Right now samples do a BETTER job at interpretation, IMO, because a human has to actually attempt to craft one into the performance. Maybe these EW/ACE products will just be - for all intents and purposes - sample libraries like we all know and maybe love, but with improved coding and transitions provided by machine learning. But we don’t know, and I am highly negative on the whole thing. I have yet to see a successful integration of AI that doesn’t sit like a heavy stone in my stomach.

Anyways, not to harp too much on it. I just see it all as a spectrum of human involvement. At one end, a real person writes with their own ideas and inspiration, and a real person interpretes it. At the other end, a prompt generates a facsimile of human creativity. In the middle, someone is removed, and someone remains, but there is still a real person there. I’d prefer the first, I do not consider the second art at all, and I am increasingly uninterested in the third. Art requires a person, and art requires effort. Otherwise, it is art in appearance only to me.

AI is also on that spectrum. "AI" is behind noise reduction on a line of noisy dialogue. Is that now-perfectly-clean line diminished because someone didn't fight for hours with clumsy tools to produce a worse result?

There will be AI instruments that essentially work in a similar but more graceful way to those today. While you argue that removes humanity, I am still not convinced that the Frankenstein job itself is imbuing very much humanity, if any. It's a necessary process to make something not sound hideous and of course that has a musical component from the person at the mouse and keyboard. But the expression and the musicality that the composer puts into that process could equally apply to an AI instrument. Whether that is through notation or riding midi controllers feels like a detail - it is executing your musical intent just as a Kontakt nki is attempting to translate your musical intent into a bunch of code, via the skill of the people who recorded and scripted the thing. It just doesn't seem inherently virtuous to me.

Anyway, we don't see it the same way and that's fine. No-one should be forced to use AI Anything any more than Daryl should be forced to use Sampled Anything. Fine composers will use both and none.

[Duncan Krummel]

I'm not sure how you can argue this historically. Beethoven has been dead since 1827.

My point with Beethoven was in regards to you saying this:

Beethoven never played a note of his scores.

To which I pointed out that, yes, in fact he did. He was a very capable performing musician. Many of the composers we still revere today were capable musicians. If you can press a few buttons and have a computer render you a human-like performance, the areas of musicianship related to performance are less likely to be pursued by other composers, and this has direct consequences for composition itself. Composition isn't just writing an instruction manual for a piece of music; it requires a depth and breadth of knowledge that is just as demanding as performance, but in different areas. To be a virtuosic pianist, you have to work extremely hard for a very long time. Composition isn't any different, but we're automating so many areas of the process that I'm left asking, "what's left to do?"

So again, I find it actually quite easy to make the argument that "no, composition and arrangement are NOT contained on the stave. That's just one part of it." Your point that both an orchestra of professionals and an orchestra of 8 year olds would render different kinds of performances from the same sheet music is proof to me that there exist aspects of interpretation and musicality that aren't just absent from the written score, but necessarily absent to allow room for the individual artists to say something organic and authentic to them. Heck, even MIDI contains more information than sheet music, but not for the better, in my opinion.

If an orchestra plays it from that day onwards the stave is the only option open to them. It was the language of music that Beethoven used to compose and arrange. One conductor and one orchestra can add their own interpretation on that of course, but from the original composer's perspective their notation was their complete intent.

The written notation is not the only thing available. Common practice of the time period, so historical and contextual knowledge; instrumentation and instrument construction of the time period; Beethoven's individual life context during the period he wrote a specific piece of knowledge; the modern-day musician's and conductor's individual interpretations, which often times include social commentary by means of which pieces are selected and what those pieces convey; where the music is being performed, both in terms of social or political context as well as venue acoustics. If I spent more time, I could think of others. None of these things are in the score, but all of them are important and often times separate the best musicians from those that don't care to put in that kind of work and research. Finally, none of us can actually comment on whether or not a score represents a composer's complete intent, so that's a moot point as far as I'm concerned. A composer's intentions don't require the performer to adhere to them, and neither the composer nor the performer(s) can police how an audience interprets the music.

The idea that composers can be more than that is a modern one.

I don't think this is true at all. From what I can tell, and mind you it's been over a decade since I took music history in undergrad and half a decade since grad school, being an effective interpreter/performer and being an effective composer only started to split in Western music in the last thousand years. Even then, the most successful and enduring composers of the last thousand years have largely been prolific performers, conductors, sometimes educators, arrangers, etc. I would say what's modern is the idea that composition is an isolated activity.

AI is also on that spectrum. "AI" is behind noise reduction on a line of noisy dialogue. Is that now-perfectly-clean line diminished because someone didn't fight for hours with clumsy tools to produce a worse result?

No, but we're comparing very different use cases of this technology. Our ears do noise reduction constantly. I'm not offended that my ears are - in a technical sense - distorting the sound waves entering them. But, cleaning up noisy dialogue and pretending to be a human performer are two separate ethical situations. If I hear denoising (or rather, don't hear noise) in a movie, no one is attempting to deceive me that the denoising is actually done by a purely human process. AI performed music is doing exactly that. A uniquely human process that's being stripped of its function, context, and meaning. I think that's extremely sad.

There will be AI instruments that essentially work in a similar but more graceful way to those today. While you argue that removes humanity, I am still not convinced that the Frankenstein job itself is imbuing very much humanity, if any.

Well, you're being hopeful with your first sentence. We don't know what it'll be. Hope is important, but I see writing on the walls that communicates a different story. But also, the Frankensteining of sounds that sampling does is also a problem! I think I've made this pretty clear, but I'm not a fan of sampling either. All of it sucks, it really does. The music I hear more and more of in media is so obviously sampled that I've largely stopped caring about media music in general. It all sucks.

But the expression and the musicality that the composer puts into that process could equally apply to an AI instrument. Whether that is through notation or riding midi controllers feels like a detail - it is executing your musical intent just as a Kontakt nki is attempting to translate your musical intent into a bunch of code, via the skill of the people who recorded and scripted the thing. It just doesn't seem inherently virtuous to me.

Disagree. All of the current AI technology exists purely to automate processes. The expression and musicality that has to be put into a sample-based mockup requires intent and effort. AI doesn't, because it tries to automate everything. So it reads to me like you're actually supporting the idea that it's the interpreting of the intent that's important here, which I agree with. But, I don't agree that AI can interpret intent at all. It doesn't, it approximates, averages, predicts, what-have you; but, it does not interpret, because interpretation of the kind we're talking about - the kind that requires real experience and perspective - requires consciousness.

Anyways, don't let me overtake the conversation too much. I appreciate the dialogue, but in real life I need to be careful about taking up too much conversational real-estate, so I'll take a little pause on this one.