The holy grail of notation programs!

So I was wondering....ideally, for a computer-using composer, there would be some maximum performance/efficiency he could reach; the sounds and playback would be so real that a live performer could essentially be cut out of the picture, and these sounds would integrate perfectly with the program, offering controls to change it simply, intuitively, and is, well, what a "holy grail" would be: perfect.

Basically: When I write in a piece and it's like I have live players, without actually having to get some live players. and these ones do exactly what you tell them to, every time ;).

I believe that this will be possible in our lifetime. It's difficult to say when, obviously, predicting the future is tricky business, but I'd guess around 15 years, with a good growth of technology and economy.

When it does come around (and believe me, all signs point to it coming ;)) it will raise interesting issues about the nature of computers versus people; the AI holy grail would be foreshadowed in that of music playback. The question of music's soul will be brought up the same way the question of peoples' souls will be when AI has our intelligence (or much more).

Fun stuff to think about, I think, but not much you can actually do about it but wait :P.

http://www.garritan.com/Strad/AngryStrad.mp3

It's not far away from the future and it really depends on what you expect. You need to know that samples are being used in film and computer games music, with amazing results already! Blended with a few live instruments, they sound perfect!

The idea is one: To enable someone withhout the budget to get the killer sounds. To enable the pianist to have guitar and violin sounds. To enable to guitarist to a last read notes (:D:D no! I'm kidding, Mark :D) To enable any composer to have a better pallete. Some libraries ARE able to deliver the final result, others not so much! Depends on what you want and what you're after. ;)

Mind you, going by your thread title, I think it is ALSO important to note that proper notation would be a HUUUUUUUUGE asset in this case.

I see so many improperly notated scores every day that it's just sad.

By "proper notation" I also don't mean "proper" in the "prim and proper" sense. There are always excellent reasons for breaking notation conventions. Unfortunately, there are more BAD reasons to do so than excellent ones. And before you start breaking conventions it's actually pretty important to understand them and why they exist in the first place.

So I was wondering....ideally, for a computer-using composer, there would be some maximum performance/efficiency he could reach; the sounds and playback would be so real that a live performer could essentially be cut out of the picture, and these sounds would integrate perfectly with the program, offering controls to change it simply, intuitively, and is, well, what a "holy grail" would be: perfect.

Basically: When I write in a piece and it's like I have live players, without actually having to get some live players. and these ones do exactly what you tell them to, every time ;).

I believe that this will be possible in our lifetime. It's difficult to say when, obviously, predicting the future is tricky business, but I'd guess around 15 years, with a good growth of technology and economy.

When it does come around (and believe me, all signs point to it coming ;)) it will raise interesting issues about the nature of computers versus people; the AI holy grail would be foreshadowed in that of music playback. The question of music's soul will be brought up the same way the question of peoples' souls will be when AI has our intelligence (or much more).

Fun stuff to think about, I think, but not much you can actually do about it but wait :P.

It's here already - you may want to check out Sibelius 5 (Sibelius home page). Sibelius allows you to write a score in the "classical" manner, entering notes onto staves. You can choose the type of score (pop/rock, classical orchestra, brass band, choir, barbershop, et cetera). Sibelius also allows for accidentals and articulations - and guess what, you can play everything back, and then it sounds exactly the way you entered notes into the score. If you hook Sibelius up to high end music libraries (eg Eastwest Symphonic Choirs, Eastwest Symphonic Choirs, www.soundsonline.com) you can create a full blasted professionally sounding right from your laptop or desktop (I am in the middle of doing that now - upgrading my hardware this week). You will have to invest in separate and very powerful hardware, think about a very fast quadcore pentium processor coupled with massive hard disks and lots of RAM, and the software is expensive, too - and you will have to figure out how to configure it all to make it work for you (takes some due diligence and patience). And yes, you still have to compose the music, the computer will not assume that task. ;-)

(I am not associated or affiliated with either Sibelius or Eastwest, by the way).

And yet again... sibelius is NOT the proper software to use sample libraries! Just because it can does not mean that it will produce decent even results. You don't have enough control in Sibelius to do a good audio work.

Sorry.

EDIT: BTW, the way I see it since 2000 or so, notation is fine and can produce excellent results. Finale and Sibelius, both are the professional software for notation and they are used for professional reasons! And they are not emulating real life, like samples do, so there is no real comparison. :)

nikolas: yes, I know samples are used professionally, and that they sound really good ;).

peter: Finale does the same thing...sounds pretty good, but any decent musician can tell the difference between that and live, I'm sure.

QCC: yes, I'm assuming that the composer knows how to notate the music, adding slurs and dynamics and articulations and all the details. I don't expect a score with just a bunch of notes to sound perfect :P.

I agree that these libraries' sounds are realistic, but to get them to playback realistically it requires a decent amount of work, usually in a sequencer to get the best control. I'm talking about simply notating the music and having it play back realistically (as in, sounding like humans are playing it). I'm talking about a notation program that takes care of everything, where we can hardly tell the difference between what's real and synthesized without an enormous amount of input. Because in today's case, it's not the computer that performs the music, it's the composer (inputting different samples, playing with dynamics and tempo, etc.), the computer simply plays it back. What I'm referring to would know automatically what to do, and be able to manipulate the samples to get the sounds just right. In 15 years or so I think we'll have this software/hardware (the hardware to run it, obviously :P) and it will be at least as accessible or affordable as software today.

I'm not saying that this will be ideal for music (I certainly prefer writing for real players), but I'm quite sure that that's what computer notation/playback programmers are aiming for, and they will reach that goal eventually :P.

....ideally, for a computer-using composer...the sounds and playback would be so real that a live performer could essentially be cut out of the picture...

This is possibly the worst potential for technology imaginable.

:O

And it is impossible, since scores are NOT designed to give exact instructions, but allow the performer to input their own personality. ;)

Yes, I know. Instead of the performer inputting his personality the composer would ;). That's what I meant when I said the program would offer easy, intuitive controls to fine tune the performance in whichever way he chooses. You might say that this is impossible, unfeasible to do with computers, but it's likely simply an issue of power. Given the rate of growth in technology it seems like within 10-15 years, or sometime relatively soon (my lifetime, I'd say) computers will be powerful enough.

Like I said before, the question of the "soul" of performance comes in. Is it something that can be calculated? I dunno, but I'm pretty sure programmers are trying to calculate it. Hell, what do you think "Human playback" is :P?

Oh, and I agree with you Robin, I was just talking about composers using only computers, who'd send their sampled music as the finished product.

But this is ALREADY possible with sequencers. And the amount of work won't be hugely reduced really, I don't think so.

Technology is a tool, not the end product.

Does no one understand what I'm saying :wacko:?

I mean a program that's everything at once! No use for sequencers :rolleyes:....and I know it's just a tool, I've only ever been talking about it as a tool :huh:.

Er, perhaps if I explained it like this:

In some amount of years there will be a notation program, one like Finale or Sibelius, where you notate the music, let's say a piece for string quartet. The computer plays it back and sounds as if it were a real quartet reading it for the first time, albeit a bit more precise (or not, perhaps they'll slightly incorporate human errors). The composer would then change aspects of tempo, articulation, dynamics, expressions, rubato, etc.

Today this is possible with sequencers, yes, but I'm simply talking about a program that combines all of that, samples, sequencing, and notation and that is easy to use, and affordable. In short: the holy grail of notation programs. Er, I should have mentioned that I was talking about notation programs :P.

Edit: yeah, I changed the title.....

is there even a song?

.....what?

Why did you ask that :huh:?

In some amount of years there will be a notation program...sounds as if it were a real quartet reading it for the first time...

The composer would then change aspects of tempo, articulation, dynamics, expressions, rubato, etc.

And you think this is a good idea?

:whistling:

Composers compose....performers perform. Composers trying to control any/all aspects of a performance is a disaster waiting to happen.

Did I ever say this was a good idea :rolleyes:? *sigh*, I merely wanted to discuss what the consequences would be, or if it's possible, I dunno. Why do people seem to misunderstand everything I say :huh:? I'm just saying that I think that's what will happen, or at least that's what they're aiming for.

I wonder why you say that it's a disaster waiting to happen, though. Certainly it's already done in sequencers, and the results are decent, at least.

I merely wanted to discuss what the consequences would be, or if it's possible.

I wonder why you say that it's a disaster waiting to happen, though. Certainly it's already done in sequencers, and the results are decent, at least.

It's done with sequencers, yes. BUT, then the composer is dealing with a drastically finite set of possible variables. The samples (I assume here, I don't know - correct me if I'm wrong) only offer a few possible sounds.

There's an impossible number of variables (plus x number of derivatives) that performers spend YEARS trying to learn about and understand - let alone master. Composers opening that can-of-worms are in for a rude surprise. There's SO MANY infinitely subtle things a performer controls - things they often aren't even aware of - that affect the performance.

I, for one, would NEVER consider trying to understand the nuances of viola bowing techniques alone - forget trying to control and perfect EVERY aspect of that instrument's performance. Now do that for every instrument in the orchestra.... A daunting task. Perhaps there'll be a decade or so where NO new music is produced - just tweaked.

:whistling:

A daunting task? Hmmm, well the composer wouldn't tweak it directly, the program would. I'm not saying he would perform each piece, I'm saying the computer would take care of the technical aspects while the composer would manipulate the more "human" aspects, musicality issues (in this situation it's more a matter of simulating the human mind rather than sound). So yes, programming it would be daunting, but taken in small to large steps over a period of years makes it very feasible. Like it or not, all signs point to computers becoming insanely powerful, and would be able to handle the small details involved with manipulating the sounds. When? I can't say when. I guess 15 years, but that's perhaps a bare minimum.

I'm not thinking in terms of today, this is essentially predicting the future, but it's not unreasonable. I don't like it anymore then you do, but I think it will happen :huh:

...well the composer wouldn't tweak it directly, the program would...take care of the technical aspects while the composer would manipulate the more "human" aspects, musicality issues

Okay...

Like it or not, all signs point to computers becoming insanely powerful, and would be able to handle the small details involved with manipulating the sounds.

I agree with you - with enough time and money, anything is possible. I'm sure, in the distant future, that a computer program will accurately simulate EVERY variable affecting a performance (from the cellists right-hand grip, to the hydration level of a clarinet reed, to the amount of condensation inside the 2nd-valve of a silver-plated euphonium held by a female with a wooden-leg).

I also think this stuff will be quite genre specific. I expect classical to be the first to go by the way-side. Especially with orchestral playing, the emphasis on accuracy and precision make a lot of that music quote easy to synthesize - hell, aren't we seeing major film scores produced solely on the computer? I give commercial classical-type music another 7-years. ;)

Pop music is another genre we're already seeing a trend away from live musicians.

Some jazz and improvised musics (I'd like to believe) are immune, and will never be replaced. The human spiritual/emotional interaction is too important.

:thumbsup:

Well, yes, I was talking purely from a stand point of completely notated music. You didn't have improv. in mind while reading my posts, did you :huh:? I would hope not, as obviously that would require much more than a notation program.

Since you brought it up, though, what about when humans are simulated? Say a computer could act exactly like a human....pass the Turing test, for starters :P. I wonder if then improv. could be incorporated.

You didn't have improv. in mind while reading my posts, did you :huh:? I would hope not, as obviously that would require much more than a notation program.

True enough...and no, I didn't mean to drag in improvisation ;)

Many jazz-type genres are stylistically resistant to your synthesis, though. Swing feel is something that eludes many serious human players, I don't think any computer will get it quite right ;) ...other aspects: articulation, phrasing, etc I expect will also be quite troublesome, and might take a while longer to perfect the programming ...

Since you brought it up, though, what about when humans are simulated? Say a computer could act exactly like a human....I wonder if then improv. could be incorporated.

I'm sure computers will "improvise" - scraggy, they probably already do. But it'll sound...off. It'll never be convincing, and will lack the essential spark that will identify it as soulful...

I'm sure computers will "improvise" - scraggy, they probably already do. But it'll sound...off. It'll never be convincing, and will lack the essential spark that will identify it as soulful...

They do improvise! The program "Band In A Box" does it quit well (for a computer anyway). It has many pre-programed styles and I guess scales and note choices to select from and then plays them back in some kind of logical order over the chords you input. It still sounds like a computer playing it but its probably the best improvising a computer can give. Its a nice program to use as a practice tool. PG Music Inc. - Band-in-a-Box, PowerTracks Pro Audio, and More...

Gotta disagree with you robin. I think its just a matter of time. Who thought in the 70's computers would ever be able to chuck out such realistic music as they do today? I doubt very many at all. It's just a matter of time, and maybe not even that much time. 10 years? 20? It's certainly not far off.

When push comes to shove, music isn't about the players playing it, it's about the sound they create. What you say is soulfoul is just a certain pattern of air vibrations that makes you feel that it is "soulful". If a computer could reproduce that sound exactly it would feel exactly the same. Maybe if you knew it was coming froma computer you would subconciously reject it, but if you didn't you would say it was just as soulful.

OK, by todays standards I agree with you. But I'm 99% sure that it WILL happen, and probably well within our lifetimes.

Gotta disagree with you robin. I think its just a matter of time. ...music isn't about the players playing it, it's about the sound they create. If a computer could reproduce that sound exactly it would feel exactly the same.

*shakes head dismissively*

Nah, I'm sure they'll get it 95-97% figured out, but that last bit - the part that makes improvised music genuine and powerful - will never be replicated synthetically.

BUT, what do I know?! Ask me again in 30-years. I'll probably ask to mow your lawn for cash. ;)

----------------------

Also, are we only considering the recording side of things? What about live performances? No one's gonna want to see Beethoven's 5th performed by three laptops... well, someone would...but not many.

:whistling:

It's interesting to think about the methodology that could potentially be used to emulate human playback (and by that, I mean real human playback, not Finale's existing technology ;)).

I'm thinking the underlying principle could work along the same lines as motion capture does for animation. Load the score for a Bach partita into a computer, set a performer down to play it, then analyse how their interpretation of each individual instruction differs from or is faithful to the written material. If you went through this process with several hundred performers, you might be able to construct a primitive library of "Baroque performance nuances". Then, it would be a simple matter of incorporating that into a sequencer, notation software package, etc.

Given these generalised nuances, it might be possible to get it to think originally with some AI.

Jazz is probably a whole different kettle of fish. ;)