Ok. I try to contribute something useful to this forum. As my english isn\'t the best I hope you will not misunderstand what I\'m saying.
Are samples real? Can they be?
It IS possible to have a real sound of an instrument on a computer, but the CPU-power that is necessary for this is not availible yet. My example is a violin. Set the violin also by other instruments.
To get a real sound you have to \"model\" the violin virtually. You have to program a kind of VSTi that knows about every physical detail of the violin. Inside the corpus and outside, the wood, the strings, the bow... I do not want to go in detail here - yeah - everything that makes a violin a violin and not a contrabass or something. The next thing is to program a player. With his abilities to play the violin, his power in his fingers, in his arm and so on. Then you can set the violinists position in a room/hall - and then you get the real sound of it. Hmmm... let\'s see what the future brings, this is nothing for today.
But you can make even a violin more realistic than it is these days. I don\'t think that the highest resolution violin is the best. I don\'t think we need more than 24-bit. Why? When today\'s Audio-CD format was released they decided a 16-bit resolution at 44.100 Hz would be enough. Now tell me that you wouldn\'t say a violin isn\'t recorded live when you listen to the 16-bit recorded CD with quality speakers. An example: Final Fantasy is the best computer graphic shown in a film yet. But nevertheless what resolution it had you would never say that Akira was played by a human actor. The human touch is still missed in those characters. The same is about samples in my opinion: You can do a violin at 32-bit with 88.200 Hz and the live recorded 16-bit CD will still sound better than your sampled violin. It\'s not the higher resolution the more realism you have.
The lack of our today\'s libriaries are more in playability and variations. Ok - we have libriaries with more than 12 layers. But is this necessary? I think so. But the layers are not all: If you play the violion on 5 notes fff you play it all time fff. But you would use only 1 layer. But a real violin sounds different on every note, even if played fff. That is what we call realism. Yeah, GOS offers up and down bowing - right. But only one down bow fff. And let me here remember you that we do not only talk about strings, also think of brass or drums. The autmatically up and down bowing is a good thing, in my opinion it could also be connected with an automatically \"note selector\". Let\'s say 3 different samples per layer and it would be fine. Don\'t blame me for this sampleproducers, i know this would be an unexecutable recording session and twice of the work now spent on a library. The price will be about something nobody wants to purchase for it. But in terms of realism...
To say \"something works great in background\" is not the right way to more realism. If you pay attention to the detail as an advanced listener these parts will sound \"unreal\" to you.
I listened several examples of a solo violin posted by users of this forum. Some have on every note of the solo a long volume fade in. That isn\'t what real soloists would do. They play the solo in one continous line. Expression is not only to volume up and down. The key to expression is the combination of volume with vibrato in my opinion. And remember here that brass can do vibrato, too.
Let me give you one more example. Recently I listened to \"Theme From Cast Away\" by Alan Silvestri. Sorry to take this out of thousands but this is the only one I remember a good example for this. There is an oboe at the beginning playing without any background. Try to extract exact one note out of that line from its start to its end. It\'s impossible. You\'ll hear also the tone played before and the one played after the note you chose. And only one person is playing oboe here. So one thing is also to the composers out there there to pay also attention to the realese of a note. Breathcontrolling is therefore a very nice thing.
And this takes me to my last point. There are already e-violins out there in the world. Why not build a vibrato-controller with just one string and a short line connected digitally with your computer? Or just connect this Midi-violin with your computer? You\'d only need one player instead a full orchestra. Example for use of e-violin: Either Lord of the Dance or Riverdance as far as I remember.
I\'m really blessed of what is possible today. But there still can be more realism. But we know nothing is perfect...
Be aware of my mistakes... Regards.
[This message has been edited by PolarBear (edited 03-27-2002).]
These are highly interesting avenues of reflexion, and I believe we will be reaching the conclusion that a programmed behaviour will always lack the emotional naturalness and intensity of a human being, no matter how sophisticated it is... Can the programmer\'s emotions be charged in the sequenced tracks? That\'s another interesting point... And most of all: would it not be more gratifying to be a musician with a violin in his hands rather than being plugged to an interface? I definitely believe so, but hey! my sampler is my sketching tool, and the opportunity that it offers me as a composer is astounding!!!
<BLOCKQUOTE><font size=\"1\" face=\"Verdana, Arial\">quote:</font><HR>Originally posted by eliam: I believe we will be reaching the conclusion that a programmed behaviour will always lack the emotional naturalness and intensity of a human being, no matter how sophisticated it is... <HR></BLOCKQUOTE>
I don\'t think so. I think The naturalness of a sound CAN be captured, but right now it isn\'t possible. At least better than it is possible yet. Think back: who told you something about samples 20 years ago? What standard do we have today? The future brings it up... How fast id we get from a 33 MHz- to a 2 GHz-CPU...
And for the understanding of an instrument it is really better to play it yourself. You even learn new things about it if you\'re playing it already 20 years. I\'m sure all musicians agree me with this. But the heck! There is no composer out there that plays every instrument availible in the world. So I have to agree with Damon, who posted this in another thread: \"The thing about expression control is practice, practice, practice.\" And this is not only valid for expression, but also for all other compositional skills.
I agree with you on every point! We don\'t know what the future holds for us... We certainly did not suspect what we\'ve had thus far, even though when we look back it seems that it was \"predictable\" in many ways...
One composer could hardly play all instruments, and even a fraction of them, but if we\'re talking about getting as close as we can to fooling ANY listener\'s hear with a virtual musician, then we surely need 100 or 1000 times the processing power that we have to generate an ensemble, and even with that, we can fool no one\'s Heart!... In the end, I persist
in thinking that there are things which we could never emulate, without passing through the process of procreating a child... We can emulate intelligence, we can emulate musical instruments, we can emulate weather patterns, to name but a few, but we can not emulate the sparkle in a happy person\'s eyes, nor can we emulate emotional sensitivity, and we cannot emulate the sunrays which give us Life. And the day we can do it, then we will have re-created a human form with the ability to receive the Sacred Fire from the Highest Universal Divine Instance, to receive the alchemical Sparkle of Life, and the human has no mastery over that, or at least very few have... The human body IS the ultimate virtual reality interface, and I don\'t think we could create a better one, even tho it might not be impossible, with some help from other Realms...
As to emulate a musical instrument by modelling every physical part of it and everything around, that sounds like something I would use everyday!!
I got carried away, sorry! I think that the next immediate steps are obvious: let\'s look at this in a 2 dimensional perspective: vertically we have the dynamics (from ppp to fff) and horizontally we have the different possible techniques and articulations. I don\'t know how many dynamics are needed to smoothly pass from almost inaudible to very loud. It could be 8, 20, 30? More than that, I guess it would complicate things more than anything else. The other vertical aspect is crescendos and decrescendos. These are tricky, because they can last from very short to very long, start from ff to fff, from p to f, from ppp to ff and so on. Add to that that they can grow (or decrease) in a straight line or in a curved one and any combination of both... Also add that they can be performed with various levels and styles of vibrato, tremolo, glissando, etc. etc. Only with what I just mentioned, there can be weeks of work just to record the sounds, as for editing...
Aside of that, we will need all kinds of bowing (let\'s talk strings!) with many choices of notes for each velocity at each pitch, so that repeated notes won\'t sound mechanical, and to leave it open to fit another sounds (on the same bowing) when one isn\'t satisfying...
So horizontally, it\'s almost infinite, because there are hundreds of different ways to create sounds with an instrument...
Does that go in the sense of where you want the discussion to go?
Hey, I don\'t want to flame anybody for writing something related, I only showed my origin intention
But do you really need all articulations sampled? Because this horizontal line, how you call it, is in fact infinte, but if do all the articulations at least 50% of this won\'t be used. Look at that \"Kyrie\" sample in SOV - once used you always know \"ahh - again SOV\".
Crescendo: A crescendo could be done with samples yet very close to real. For example a long crescendo from pp to ff You take the p samples, fade from ppp to p, then fade to a mf sample which slides from p to mf and then fade to a ff sample which slides from mf to ff. It\'s a bit tricky to do, but it works great. And it doesn\'t sound more unnatural than the samples are. Thomes_J is using a cresendo from p to f with only change the volume in his free bamboo-flute demo (at 0:30 - 0:33).
And what I want for the articulations is just a good tool. Tremolo, vibrato or whatever - these CAN be captured with a tool and be used with \"normal\" samples, again you have to know how to use it. Everybody could learn about reverberating, why not about this, too? It isn\'t that difficult it seems.
[This message has been edited by PolarBear (edited 03-29-2002).]
I\'m not certain that all lenghts of crescendos could sound natural through this crossfading technique, but it can certainly be a good option in many cases...
Indeed, the more we will have a varied articulation palette, the least proportion of all the sounds will end up being used, that\'s a fact, and that\'s also what could make a track sound nice and organic, because of the wide variety of sounds available.
For sure my projections are a bit extreme, but if we push the limits of sampler-based possibilities, then I think that we will see more and more this type of extremely complete (maybe overly) kinds of sample libraries, provided that we can have efficient interfaces permitting the programmer to easily choose which samples are appropriate. It\'s also about new kinds of softwares which could emulate certain aspects of a given instrument\'s playing and leave the final polishing to the artist... I\'m extrapolating here, but to efficiently use a library with 20 layers and 50 articulations, we will need tools to help us find our way through it...
BTW, I never felt flamed in any way and was happy althrough this thread!!!
Here\'s something I discovered from using the virtual modeling capability of the Yamaha EX5. The more realistic the model the more you have to be able to play like a real player. Add to that that there really aren\'t very many truly convincing models. The EX5 saxes are pretty good for sounding like Kenny G, but they won\'t grunt and growl like the real thing. The trumpet model will flub a note, but getting a soft round sound is impossible. Frankly the models are still far from realistic enough for me and when they are I\'ll have to spend all my time practicing and give up composing all together.
So having essentially giving up on modeling I\'m back to sampling. Polar Bear asked, \"Are samples real? Can they be?\" I would answer that it\'s a matter of degree. I have GOS, it isn\'t perfect, but it\'s gotten me a lot closer to convincing realism than I was 4 months ago. Could it be better? Certainly! But I can make convincing music with it. See my other thread on this forum for an example.
The bottom line is there are only so many hours in a day. With my recently acquired sample libraries I have enough to keep my otherwise busy life, totally packed for quite some time, just learning how to use them. There\'s not enough time to perfectly emulate an orchestra. With samples we can get close, hopefully that will allow us to write music that someday will be performed by a real orchestra. For me I need to concentrate on what I can do and not worry about what I can\'t do.
Well, pantonality has something I\'m verry interested in, the EX5. Unfortunately it is something I don\'t want to invest 3000 bucks in, thinking of its lack in speed (disk-drive speeded scsi-interface...), and also only 8MB flash memory are possible in it (not to be unfair, 64MB RAM is also possible, but you have to uplaod all again if you turn off the system - with that disk-drive speed ). Nevertheless I like it someway. I don\'t know anything about if all the EX5-synthesis (right term?) is already availible for a computer.
Ok I\'m getting away... I really like the things that you can do with that station. But I think it is not usable in an orchestral work nor to emulate/model any acoustic instrument.
pantonality, composing is one thing. I wonder how people could compose for real orchestras only using paper and pencil. They can imagine the right tones. Based on what? They heard something (here are real instruments meant) and repeat this in their brain. Unless it is complete or not, in you\'re brain it would sound real. What is the differnce to sampling? If we can catch that sound in the brain, would we have a real sound? I can\'t answer this question myself unfortunately. I\'m not a genius.
But something I can do is think of a complete orchestra playing great melodies. But I can\'t focus one single instrument without loosing others. And I can\'t remember what I thought about. Too complex So I\'m not a paper/pencil worker...
That leads me to the opinion that I only remember complete passages I heard somewhere and create my own sound out of it. Isn\'t this sampling? Or used sampled things?