As I discussed in last month’s column with Bruce Miller, the likelihood of a composer having an orchestra record his/her score is sadly smaller than 20 years ago. Most will have to create their scores with a small group of musicians augmented by sample libraries and virtual instruments or only the latter.
So we need to learn how to do this well. The good news is that while they are not yet (and I hope never will be) as good as the real thing, in recent years some pretty dramatic progress has been made.
How we approach the craft of composing with these tools is somewhat determined by our philosophy and that of our clients. The reality is that some parts we conceive of in our mind and ears that would sound stunning with a real orchestra may not sound good with the libraries we own or any available . Conversely, we may be able to conceive of parts with the samples that real players simply cannot do. For instance, with a slide trombone playing the B flat a 9th below middle C to the B natural a half step up means going from 7th position to 1st position so there will always be a gliss. I do not know however of a single library that has that built in limitation. Harpists have to deal with certain lines of notes by pedaling enharmonically as no library requires.
So what is the best way to approach all this? There are essentially 2 prevalent schools of thought:
1. Write the music that you hear in your head and if your libraries cannot quite do it, seek additional libraries (or bitch and moan about the inadequacies of even the great libraries of today:) I see this approach sometimes advocated on forums.
2. Don’t worry about what the real guys may play, write what sounds best on the samples, unless you know you are replacing the parts with real players, at which point it is more or less a non-issue other than for the client to hear it before the recording session.
Or is there perhaps (timpani roll) a THIRD way! :
My approach has always been a third way that leans more toward number 2. For me real, shmeal, none of it sounds real when you put it next to the real thing, so if writing something differently or eliminating something that you would write if it were real players because it does not sound good with the samples, then so be it. That said, it is always in my mind what the real players do. You break the rules better when you know them.
Back when I was scoring “Zorro” I used to add a Memorymoog to the real celli parts to fatten it out, along with some E4 samples. Did the end result sound more “real?” No. Did it sound better? To my ears, yes.
However, as I have discovered, my view is not universally accepted. So I reached out to some colleagues to get their take on it and found some really interesting differences in point of view and yet, surprisingly, a sort of consensus seems to emerge.
John Frizzell advised, ”Take an existing ‘great work’ composed by one of ‘the masters’ and sequence it (choose something in the ballpark of what you plan to write) then import an actual recording of this ‘great work’ and match it as best you can, including tempo. IF your sequenced version of the ‘great work’ sounds like a ‘really good work’ then your samples are ‘really good’. If it sounds like crap, your samples are crap and you know that if you compose your own ‘master work’ it too will sound like crap even though it might be a ‘master work’. I have sequenced many many ‘master works’ and some portions of a pieces sound incredibly real (with great samples) and some sections sound like crap with the best samples money can buy. This is a constantly evolving thing as new libraries and technologies emerge.”
C.J. Vanston replied to me: “I write what’s in my head, the sample libraries are not as important as the way I play the stuff in live, using volume pedal, pitch/mod, basically a live performance. My libraries have evolved over the years and are more than good enough to express what I’m trying to say. And I’ve been emulating live instruments since a Prophet 5, coaxing emotion out of less than great sounds. I’ll take that any day over some guy with the great sound library who draws in all the expression, volume, vibrato, etc. but that’s just me. My live performances are a huge part of my sound.”
From Michael Levine: “The third way: write for your samples to sell it to the client then fix it on the page before the real guys play it. Not so often with voicings and instrument choices but often with articulation, note lengths, and effects on the instrument. But you can get bitten – many times I have replaced samples with more exciting real performances only to be told to go back to the demo. There are also demo “tricks” I use like doubling string groups with synth pads for smooth transitions in demos that are unnecessary once you have a few dozen individuals making the same move at slightly different times.”
Nathan Furst was pretty much in accord with Michael: “3 (ish) – write what’s in your head, and when the samples make that difficult/impossible, adjust for the strength of the library and move on. Not unlike live players- sometimes they can’t pull it off, so you make changes and move on. “
John Graham had a response that was..err… colorful! “I think there is a third way, though it may really be number 2 with a half rotation. If it exists, the third way would be this: Write what you hear but be willing (as I just did on a score) to drop, say, the clarinet sample if the flute samples sound better on the line, even though a real clarinet would sound better and that’s WHAT I BLOODY WANTED, Y’BASTARDS!!! And that’s even if you are replacing sounds with live players later. I never, never, never hand over even a demo of a piece unless I think it sounds practically good enough to put into the film even without the live players. If that means substituting one instrument for another, I do it. If that means that I have, in effect, 200 strings or 18 French Horns playing a passage, I care not.
Craig Sharmat’s view is somewhat mercurial: “Yes there is, at least for me…and that is write what is in my head and sometimes adjust to the samples…or write what’s best and realize if you take what guys would play into consideration, start altering your parts so they are more realistic. I come at it either way depending on how I feel inspired that day.”
Kays Alatrakchi’s take: “I just completed scoring an animated project very much related to the Star Wars universe which required an original, but heavily leaning towards John Williams approach. The way I compose in those instances might very well be the “third” way you’re mentioning. Essentially I start out to write the music that I hear in my head, however as I pull up samples to realize the various parts of the score, I am forced to change and adapt my original vision to fit what sounds best and what the samples can handle. I suppose you could describe this as a hybrid between method 1 and 2? I also think it’s important to do things while writing with samples that would not be done with a real orchestra. For example, I’m bound to play the trombones in octaves because I like to layer the crack of the low end with the definition of the high end. One of the things I do very often is also add “noise” to the background…this is activity in samples that in many cases might be playing “jibberish” but that add to that sound of 80+ players doing stuff. These are all things that in real life I would never do, but that I find necessary when working with samples. There is another trick that I have used to great effect that I think many composers would find extremely controversial…so I’m not sure if I should even mention. The good news is that due to the added capabilities of sample libraries, it is becoming more and more feasible to create what’s in one’s head without having to adapt or resort to cheap tricks.”
Peter Schwartz raised a great point: “Unless a composer conceives their music solely in their head and writes all of their ideas down on score paper before embarking on a mockup, it’s inevitable that the samples will shape the sound of their music. The degree to which this occurs depends on the libraries they have and what kinds of emotion they allow the composer to express. In a talk with Bruce Broughton the other day, he was of the opinion that his score for Silverado wouldn’t have sounded the same were he to have had to mock it up first, for the type of emotion he wanted to portray just wouldn’t happen with samples.”
Andrew Keresztes, (who is also the developer of the highly regarded LA Scoring Strings library) pretty much came down where I do: “The answer is fully predicated upon the end-use of the mockup. If it’s for commercial use (as opposed for demos) then you cater the composition to the strengths of the library and use what you have to the best of its abilities. If it doesn’t “sound” good… then don’t use it… I find that shoving a square peg in a round hole rarely makes the piece work.
Tom Salta was succinct: “Coming from more of a record producing background, I usually use #2 with a little #1 sprinkled in!”
As was Ray Colcord: “I’ve always written towards the strengths of the samples I have. I’ve got a bunch of libraries, but there are some articulations, especially in strings, that the samples just don’t do well, so why swim upstream?”
So, how little or how much to be guided by “reality?” Aye, there’s the rub. For those who compose “concert hall’’ music who think it unlikely that they will ever have an orchestra or those film/TV composers who have clients who demand that it must sound “just like the real thing” well, they are going to move heaven and earth to bend the samples and virtual instruments to their will, and that is understandable.
But for the rest of us, here is where I believe the general consensus falls:
Knowledge of what the real players do is very helpful and should guide you but should not shackle you.
Every respondent agreed that adding even a few good quality real players when possible usually helps with both believability and quality.
I would add this final thought: try to enjoy the process. Don’t make the elusive attempt at “hyper-reality” turn it into drudgery. Yes, it is hard work, but that does not mean we must suffer.