Creating realistic string parts with samples is an art in itself. While it is necessary to have at least one quality library, the lines quickly blur between traditional orchestration and modern sequencing. Knowing when to emulate the time-honored techniques diligently —and when to ignore them and exploit the virtual aspect of virtual instruments—is oftentimes the key to effectiveness.
Fig. 1. Cinematic Strings loads its entire collection of long and short notes as key-switched articulations. A unique feature is a knob that changes samples for playing high or low on the neck. Nothing is more boring than strings playing long, sustained notes the whole time, which is tempting when pulling up a string sound on a keyboard. It’s a good idea to get familiar with other articulations, especially short bows, as they’re almost always necessary alongside sustained notes to pull off authentic melody lines. Make sure a good articulation collection is loaded up ahead of time, so that it is easy to switch between them as necessary (see Figure 1).
If strings are not carrying the primary melody (as in a pop arrangement behind a singer), try to find the empty spaces in the music to highlight lines, accents and color. Think melodically, especially for the violins on top and celli on bottom, then fill in the middle parts as necessary. The lines themselves don’t have to be brilliant compositions; however, a well-shaped simple melody you can sing that fits into the arrangement will do wonders for making samples sound better.
This also applies to frequency ranges. Try to find the “holes” that aren’t being filled and voicings that aren’t being played by other instruments. Strings sound great in open voicings (notes of a chord spread far apart), so use that to your advantage. Less is often better—you don’t want to constantly double notes already taken by keyboards or guitars and thereby clutter certain frequency ranges. Samples tend to cause muddiness a lot faster than live instruments, and be especially aware if a vocal is present. One technique that works particularly well for sampled strings is to double a melodic line two octaves apart, omitting the octave in between.
Fig. 2. Dan Dean Solo Strings contains some great small ensemble patches that change the number of players via key-switches or modulation wheel. When time permits, I often build up each section of the strings with multiple passes of smaller ensembles from different libraries. When going for a big orchestral sound, I usually settle on a main ensemble patch from the library that most closely suits the character I’m looking for (that often changes with each piece). If that particular library also contains some small ensemble or solo instruments, I may also grab one of those for uniformity. More times than not, I usually turn to my other libraries for their small sections and solo instruments, as I’m also looking for some contrast to the overall sonics with these added layers (see Figure 2). Especially with short bows, the feel is usually different enough between libraries to add some nice complexity that sounds more human.
Once I have a handful of patches for, say, the first violins, I record each pass as a separate MIDI performance and pay close attention to each patch’s unique strengths. Here, MIDI control messages really come in to play. The concept is simple: Make sure the best attributes of a particular patch shines, and use them to mask any weaknesses in the other layers at any given moment.
Fig. 3. In this fairly simple sordino arrangement, two sample layers are used for each section. The controller information for dynamics and volume is unique for each patch, with the controller data for L.A. Scoring Strings on its own track in Logic. Each library administers their controllers differently, but make sure you have separate authority over CC 11 (expression) for riding the dynamics of the line versus CC 7 (volume) for setting the overall balance (see Figure 3). In reality, I usually have to edit CC 11 after all my parts are initially recorded to get both the layer blends and overall melodic shape just right; later, it’s not uncommon to automate CC 7 for more tweaks when mixing. Some patches let you modulate attack and release times (CCs 72 and 73), while others address envelope in the library interface.
Due to the relative uniformity of the string timbre, you don’t have to be quite as finicky about maintaining the correct number of players at all times as with woodwinds and brass; however, keeping the “weight” of the sections under some control when sequencing goes a long way towards maintaining authenticity. Be especially aware during divisi writing, and don’t be afraid to omit layers from time to time.
Unlike many keyboard instruments, the intonation of a live string section is not equal-tempered. String players instinctually and dynamically adjust their pitch to more pure intervals based on the harmonic mission of the moment. A general result is that lower notes are slightly flatter than equal temperament while higher notes tend to rise a bit sharp, as the resulting intervals are actually more pure. If your workstation or sample library allows, setting your string patches to “stretch” or “pure” can make them sound more realistic in certain settings; especially alongside an acoustic piano that has been stretch tuned.
Sometimes that last bit of magic isn’t found in your sample libraries at all. A common technique among jingle and TV composers is to add a live player on top of the string samples. Most common is to record a single violin and/or a cello on top of the samples, which does wonders to smooth out all the transitions between articulations. A good player will intuitively connect all the phrases in a refined way, typically making your samples sound more realistic.
Be sure to not record too closely, as you want to complement the samples, not stick out like a solo. If you can get three passes out of your live player, all the better. Repositioning the microphone between takes and using some subtractive EQ to remove unwanted frequencies should help to further blend them with your samples, especially after applying a unifying reverb to both.
In lieu of a live player, an old trick is doubling a high line with a string sound from an analog synth, which can provide a certain presence and life. Tweak the filter cutoff slightly towards the bright side, get rid of any built-in chorus, set the attack and release fairly short, and tuck it in until you only notice its absence when muted. You’re listening for air and a little edge, not more mush.