Article: Let the samples dictate the music

[Mattias Westlund]

New article. This is something I've been thinking about for a loooong time but I'm still not sure whether I've managed to get my point across in a way that makes sense. It turned out somewhat ranty. I might revise it if/when I find the words, but this will have to do for now.

[Paul Battersby]

I read the article. I think you presented your point well and I've certainly experienced taking a library outside it's comfort range and needing to seek a different library for that one specific style. I was using the brass from an orchestral library in a stage band context. Didn't sound right until I tried brass from another source more suited for the style.

In your article you mention composing to suit the available articulations of the library but do you have any advice on mixing libraries? I'm not just talking about strings from one library and brass from another (though tips for that would be welcomed too) but suppose I need sustain strings from one library and staccato strings from another. What can I do to best ensure it sounds like the same player on the same instrument recorded in the same environment? Is it just a matter of playing with EQ and reverb? I would imagine you had to deal with this when piecing together the samples for your SSO library.

[Mattias Westlund]

(07-04-2016, 02:13 PM)pbattersby Wrote: Is it just a matter of playing with EQ and reverb? I would imagine you had to deal with this when piecing together the samples for your SSO library.

Panning, EQ and ambience/reverb play large parts, yes. Panning for making sure all articulations originate from the exact same point, EQ for making their timbre more similar, and ambience (if necessary) to make close samples sound more roomy.

I like my compositions fairly wet so I can get away with combining things quite easily, but it's harder the more intimate you want everything to sound.

[Nayrb]

(07-04-2016, 03:37 PM)Mattias Westlund Wrote:

(07-04-2016, 02:13 PM)pbattersby Wrote: Is it just a matter of playing with EQ and reverb? I would imagine you had to deal with this when piecing together the samples for your SSO library.

Panning, EQ and ambience/reverb play large parts, yes. Panning for making sure all articulations originate from the exact same point, EQ for making their timbre more similar, and ambience (if necessary) to make close samples sound more roomy.

I like my compositions fairly wet so I can get away with combining things quite easily, but it's harder the more intimate you want everything to sound.

Is it safe to assume these considerations apply to section building as well? As in using different samples of solo instruments to create sections.

[Mattias Westlund]

(07-04-2016, 03:46 PM)Nayrb Wrote: Is it safe to assume these considerations apply to section building as well? As in using different samples of solo instruments to create sections.

Kind of, yeah. Except that you would use panning for creating a suitable stereo spread to the section rather than placing all players at a certain L/R point.

I'm not really very scientific about creating fake sections though, it's more about experimentation than having a fixed set of rules to follow. Sometimes solo samples can be layered and sound wonderful together with very little effort. Sometimes they will sound like ass no matter what you do. It all depends on the samples.

[Mattias Westlund]

I might also mention that there's no ancient, arcane magic involved in the creation of fake sections like I did with SSO. Anyone can do it. Fire up a multitimbral sampler, load a couple of solo instruments into it, set them to play back on the same midi channel, and you're well on your way. Pan them a bit to the sides, balance their levels, listen to how they work together. Tweak where necessary with envelopes, filters and possibly EQ (i.e. route them to separate audio outputs). When you have something that sounds decent, add more instruments. Lather, rinse and repeat.

Just wanted to make this clear as questions like this pop up from time to time and it seems most people think it's more difficult than it actually is.

Well, coming up with something that sounds good IS difficult and you will fail 95% of the time, as did I. But the purely practical aspects aren't difficult to learn.

[Nayrb]

(07-04-2016, 03:55 PM)Mattias Westlund Wrote:

(07-04-2016, 03:46 PM)Nayrb Wrote: Is it safe to assume these considerations apply to section building as well? As in using different samples of solo instruments to create sections.

Kind of, yeah. Except that you would use panning for creating a suitable stereo spread to the section rather than placing all players at a certain L/R point.

I'm not really very scientific about creating fake sections though, it's more about experimentation than having a fixed set of rules to follow. Sometimes solo samples can be layered and sound wonderful together with very little effort. Sometimes they will sound like ass no matter what you do. It all depends on the samples.

(07-04-2016, 06:07 PM)Mattias Westlund Wrote: I might also mention that there's no ancient, arcane magic involved in the creation of fake sections like I did with SSO. Anyone can do it. Fire up a multitimbral sampler, load a couple of solo instruments into it, set them to play back on the same midi channel, and you're well on your way. Pan them a bit to the sides, balance their levels, listen to how they work together. Tweak where necessary with envelopes, filters and possibly EQ (i.e. route them to separate audio outputs). When you have something that sounds decent, add more instruments. Lather, rinse and repeat.

Just wanted to make this clear as questions like this pop up from time to time and it seems most people think it's more difficult than it actually is.

Well, coming up with something that sounds good IS difficult and you will fail 95% of the time, as did I. But the purely practical aspects aren't difficult to learn.

Oh, definitely. I had in mind something a little more simple though, specifically woodwind sections (so like two of the same instrument and not sixteen). I guess it's more like layering than section building in this context. I was musing on it because I realized that, with the exception of the SSO patches, I don't have any ensemble woodwinds (x2 flutes, etc.), and most of my woodwind samples seem to be more of the intimate / close miked variety. Not sure how practical I'd find it in the long run, but it seemed that the small size of such a section would allow for some flexibility in terms of divisi lines and such.

[Mattias Westlund]

(07-04-2016, 07:39 PM)Nayrb Wrote: Oh, definitely. I had in mind something a little more simple though, specifically woodwind sections (so like two of the same instrument and not sixteen). I guess it's more like layering than section building in this context. I was musing on it because I realized that, with the exception of the SSO patches, I don't have any ensemble woodwinds (x2 flutes, etc.), and most of my woodwind samples seem to be more of the intimate / close miked variety. Not sure how practical I'd find it in the long run, but it seemed that the small size of such a section would allow for some flexibility in terms of divisi lines and such.

YMMV as always, but personally I wouldn't bother with creating baked woodwind sections from solo samples. Wind sections are small and take well to layering as long as you make sure the tuning is spot on, so having (e.g.) a sampled three piece flute section gives you no real benefit over three individual solo flutes that you just overdub at the midi stage. In fact I think three separate flutes are more useful as you can easily move between unison playing and harmonies.

[Samulis]

Thanks for this article, Mattias. This is something I definitely preach, especially to people writing demo tracks. There's a very particular sound that each library desires- for EWQL, it is that line between that old mid-2000s-flash-game-'adventure'-sound and mid-2000s-film-scoring-chugga-chugga-cellos-sound, for VS CO, it's small, tension-rich (6's, 7's, 9's, 11's) subdued material with an almost ambient sound. When people try to force a library, it is painfully clear.

To me, it is simply counter-productive and foolish to think in ANY other way than using the instruments you have when composing digital music not intended for a real performance. Use the samples, not the instruments. Many are unwilling to change the 'image' they have for a piece, even if it means making it sound much much better with the instruments they have; to this I say, Neanderthals were unwilling to change their hunting habits when early Homo Sapiens came around, and we don't see any of those guys anymore.

When writing for concert music, I use default soundfonts/sound-canvas sounds, because I don't care what it sounds like, all I am checking is whether harmonies sound correct.

[kmlandre]

Samulis wrote:

"Many are unwilling to change the 'image' they have for a piece, even if it means making it sound much much better with the instruments they have..."

Wise words.

Which is why I contend it's better to approach a piece with as few preconceptions a possible. Maybe a melody and/ or a progression, but little else. Save orchestration for when you can actually hear what you've got to play with, not what you think you hear in your head.

And start with the smallest template possible and add in only what you need. Like a toddler with a box of 128 crayons, "I can haz all teh orch-stra!" templates tend to produce the audiological equivalent of the of the color brow.

At least in my less than humble opinion... ;-)

-- Kurt