Here’s part ii of Christian’s take on “how to program realistic sounding strings” this time around, ‘Symphonics”. With some insights into the more technical engineering aspects of working with sampled orchestral material. Here’s the shooting script Christian worked from, but to see these principles in action make sure you watch the movie above!
Hello my name is Christian Henson and aside from my work with Spitfire Audio I also moonlight as a film, TV and Games composer. I’ve done a bunch of orchestral scores but there’s just one catch. I can’t really read music, I don’t know much theory and I didn’t study this at college. So here’s the second part of my tutorial on how to program realistic strings.
Last time we looked at chamber strings which we established is not in fact a form of music for sex pests, but merely refers to an orchestral band that is smaller than a symphony orchestra. Where strings are concerned I would say anything between a couple of players, to around 25 strings players would qualify as Chamber. We discussed that any string band is essentially five voices, and that violins can play high or low, but must always play above the violas, cellos and basses. We suggest these rules not because they can’t be broken, it’s just when you’re trying to create realistic mockups of real instruments the first and easiest rule is to empathise with what they can do. This will restrict your options in a way that channels you toward reality.
The other thing I’d recommend if you want to program realistic string parts, is listen to good string music. Immersion in great music will help the us non-trained individuals how to instinctualise good writing & arranging as opposed to studying it. That is, before you hype the bejesus out of it.
So in this tutorial we’re going to look at Symphonic strings. In many ways these are easier to program than chambers, so I’m going to hold our concentration by complicating matters in this episode with some technical tips, which we’ll get to in a minute.
A symphonic string band typically is 16,14,12,10,8 a total of 60 players, but for me I would say anything above 8 1st violins and you’re getting there.
For me there’s three different type of string arranging, I call them the three C’s:
Which is when the strings play like a choir, a big mass of chords, moving together pretty much at the same time.
It’s when different sections or voices talk to each other, whether it be in a polite question or answer, or a ruder argument where both voices are pushing in different directions and sometimes at different times. Often referred to as counterpoint. Imagine two people speaking over each other at the same time, using different pitches so they can get their voice heard.
This is when a string band is a bed or a bubbling rhythmic underscore for another voice or instrument.
Different string sizes tend to focus on these different classes of orchestration. A quartet rarely underscores and accompanies, as it’s the only band in the room. It is very often conversational and occasionally choral. A chamber band is most versatile and will perform all functions. But doesn’t sound as awesome as a symphony band when it comes to choral. A symphony band can be cumbersome in conversation but awesome as an accompanist. All these are vast generalisations, but may help you when you’re picking which sized sample library you’re going to work with.
Returning back to my chamber strings tutorial, it is always good to remember that you have five voices to play with. This is especially true of Symphonic samples. Go beyond 5 voices and the whole thing can start to sound like an awful string patch on a synth. You can split the sections up in a symphony orchestra so each section plays more than one note. But what I would do with programming is add welly and weight to each voice so you get that enormous sound, but try and keep it to five voices so you don’t create a thick condensed floury gravy.
The other thing to remember with Symphonics is that each section doesn’t rely on the next to produce harmonic support and sonic richness. The numbers are so big with a Symphony band that each section has a silky richness to itself. This means you can really spread the strings across the range, from deep down in the basses to in the gods with the violins. So… as a vast generalisation, these kind of finger shapes work excellently for big bands and not so good for small. I wouldn’t do this with Symphonics unless you’re going for a synth type sound.
I threatened some technical stuff, but think it wise to crack on for a bit. So here goes, a new piece sequenced as explained in my last film, using a keyboard controller, logic as my DAW and a fader controller with expression, modulations, volume, and vibrato assigned to channels 1 through 4 respectively. Remember, keep these moving no matter what otherwise it will sound like you’re conducting an orchestra of cadavers.
I’m using Spitfire Symphonic Strings, which is also known as Mural. And like on my chamber tutorial, the majority of what I use are ensembles patches — which are baked presets using all instruments in a patch that you can play with both hands — with single sections to add detail, focus and definition. We can select articulations which are kind of like vowels with these switches on the interface, or via key-switches on the controller. But I like the old fashion way of having one articulation per track.
As I mentioned before, programming is like an oil canvas, so the first thing I’ll do is sketch the whole thing out. I do this because this is always the last thing I do on any given day. Its when I’m most lubed up, in gear and not struggling with imposter syndrome. This really is my job, I’ve been at it all day, let’s give the small guy who has after a night’s troubled sleep, fully reset and believes he cannot possibly do this for a living, a little something to work with. For me the thing that stops me getting out of bed in the morning is a blank screen. If your charcoal sketch is already there, all you have to do is think about what you’re going to do with it.
OK, with the charcoal sketch now done let’s take a brief technical interlude. Other than “what controller do you use?” the most asked question we get here is which reverb we use. Or “what is the best reverb?”. We’ll get to that in a minute. But first about these Spitfire string samples. We recorded these just as you would any string band in Air studios. So you’re ostensibly working with a carefully created board mix where each set of mics is given a Stem which stands for Stereo Master. So you can mix between them like this. So here’s the close mics, which are spot mics a few feet away from each desk of the band (a desk by the way represents two players who share the same music on the same music stand). These spot mics have a very modern sound as they started to be introduced as mixing desks got bigger. They are great for intense and tight action sequences and for focus and crispness. The next is what I tend to only use is the Tree mic which are three mics in a decca formation placed above the conductors head. It’s kind of what he hears although you hear more of the room and the players at the back because they tend to be around 8 feet off the ground. The next are the outriggers which are just the same height as the tree but are placed at equidistant medians between the outer edges of the band. These have a rich full and blended sound but are nice and wide. Finally we have ambient mics which sit way high up, around 12-14 feet up to really capture the room reverberance. These are great for surround feeds, but we won’t go there today.
Some sample libraries really need reverb to bring to life, because they’ve been recorded dry, or close, or not in situ. Spitfire libraries don’t theoretically need reverb because they have plenty of room options. But the truth of the matter is, most engineers handed a desk mix like the one offered up by Spitfire’s libraries would still add a bit of splosh to flatten and merge the whole thing together. To make it sound super widescreen and cinematic. There is also one very important consideration when using samples and that is, if they are well produced, they’ll be performances. If they’re performances it means there will be some inherent issues that you’ll need to fix with reverb.
Namely longs versus short. Listen here to some of just the tree mics for our short articulations? And now our long? You’ll notice the release triggers (which are samples that are triggered when you let go of a long note) on the longs don’t sound as reverberant as the shorts, listen again. Now this isn’t because the shorts are louder or create more reverberation. It’s not because Spitfire Audio have programmed it wrong. It’s simply because when you ask a string player to stop playing she or he doesn’t stop dead, they gently bevel the end of a note off by doing something called a diminuendo. Imagine sitting in the back of a cab and you ask the driver to stop. Imagine then if he were to hit the brakes with everything he’d got, you’d not appreciate that experience. So the same is true with orchestral players, they’ll let you off gently. The net result is release triggers are always quieter for longs, but this is the natural sound. The way to fix this is by stemming your strings into different signal paths which can be effected differently with different reverb levels. There are three reasons this is a great idea.
OK so what is the best reverb? I can’t answer that other than “the most expensive”. But there are two important considerations.
For me this triangulates choice of reverb to two or three choices. Point one is fairly straight forward. Pros use TC6000s above everything else. It’s a massively expensive piece of outboard kit that costs about 10,000 USD. TC electronic have put out different iterations of the TC6000’s VSS3 algorithm, as part of the discontinued power core range for example. These are worth looking into, and I’ve heard rumours that you may hear the words VSS3 and Native soon. If you do and you can afford, buy.
The second is working with samples, and this leads to a hot topic, impulse response versus algorithmic. By the nature of working with samples you’re often using the same moment in time, over and over and over again and when you’re writing you’re playing back, correcting and mixing the same moment in time over and over and over again ad nauseum. So if you combine this with yet another sample technology, impulse response or convolution based reverbs, you’ll be triggering one moment in time by another, over and over and over again. Now I’m not saying convo or IR reverbs aren’t good. They’re all over final mixes of my music, they sound good. But when I’m writing, I just don’t feel happy using them. Because every time I hit that response, I’m triggering that same sample. So, for me, it’s algorithmic all the way. Which the TC6000 is but also the lexicon reverbs which you’ll see around studios. In conclusion:
I’m using the random hall here, and have matched the TC6000 go to patch, the warm hall, with 2.6s delay. Now here’s the best stemming configuration I’ve evolved into using over the years. For now I’ll just dwell on why it works from a unified wetness point of view. So I stem into four string groups:
The reverb return is on 100% wet, we don’t want any live signal going through that as it will phase with the live signal in the busses or auxiliary channels. Then I’ll add a bunch of splosh to 1, less to 2. very little, or just a tickle to 3. And the most to 4.
Once you’re happy with how these are set, you can then adjust the total reverb level to how wet you want the mix to sound. I find you’ll add more the more tired you are and the later it is in the night. Kind of like when mixing vocals, you already know the lyrics so you think you can hear them. Fact is they may have become buried. Same applies to reverb, you think you can hear every moving voice in the strings, but in fact you’re imagining it, because you know it’s there. I’ll come back to some more techniques when we get further down the line with this composition.
Ok so I’m happy with this, before signing off, I’ll just have a listen and pull the reverb back. After a while of listening it will be refreshing to hear it drier, clearer and I’ll gently pull the return fader up to a point that sits nicely with what I’ve grown used to, and this nice less soupy sound. Finally if you’re preparing for a session with a live player or group you’ll find you have an excellent set of stems to tracklay and work with. For me, live strings are all about the lyrical articulations, so if you’re going to blend live and fake, you’ll really want control over the longs and live. Conversely you may be disappointed with the tightness of real players in a stressful recording situation, so you may want to blend short staccato patches so they’re just a little shimmering crust above the surface. But most importantly if you’re working with restricted numbers, really consider how you use your forces for textural elements. So where I’ve added those harmonics two octaves up, or little held shimmering tremolandi, you want to divide your sections up to play those things where the samples are perfectly believable. And finally, pizzicato. These can be a world of pain for the player and mess with your expectations entirely. They will be looser than you imagined. Don’t be afraid to do a back up take where you ask the band to “tacet all pizzicato” believe me, if you’re using good samples, those will work just fine. And finally if you’re planning to 100% use a real band, these stems may give you the back up and security you need just in case something was missed. You don’t need to dial in the whole sample band, but an awry bass pizz or slightly fruity harmonic can always be fixed at a later date.
Next time solo strings, vibrato and quantisation. Thanks for listening, here’s the cue…