The Sample Spiccato Conundrum

I've talked on this platform quite extensively in several posts already about the fact how technology shapes media music. This goes even as far as changing the expectations about what an ensemble can do in comparison to their sampled counterparts.

And I'm not talking about the uninformed newbie in the field who just bangs on their keyboard in ways a real orchestra musician could never play it. I'm talking about specific instances where what we're used from samples has shifted the way of what our clients or even we expect from real musicians.

A prime example in this regard is the very popular string spiccato. We all know the tracks that consist of aggressive spiccato chords or ostinatos with an epic horn or violin line on top.

But let's dig a bit deeper here:

In the historic development of orchestral samples, the first few generations of libraries had sustains and short notes. Sampling these notes happened in isolated context. This means the players in these sessions had a score sheet in front of them where each note was written as isolated event to be recorded. With the advancement of sampling technology, for a long time this didn't change until some developers came to the conclusion that contextual samples sound more real, so they started recording these notes in musical context, e.g. repeating one spiccato note several times and picking some out of these to later make it into the sample patch.

However, there is one specific limitation in most cases which is the fact that the players would need to leave gaps between the notes so they don't bleed into each other, which would make them useless especially in a context of changing different notes. However that gap causes a fundamentally different approach in the production of these notes by the players as they still become more of an isolated event again. In a rapid succession of spiccato notes the bow starts to "bounce" on the strings which does not happen in the same way when you leave space between the notes, so the sound changes.

Also, approaching an isolated spiccato note for a player allows them to play it way more forcefully and aggressively than they would ever be able to if they were playing a rapid sequence of spiccato notes.

However, these isolated notes made it into the sample patches as the top dynamic layer and this is where they changed the perception that many composers have of string spiccatos. The technical possibility to play these heavy force isolated spiccatos in rapid succession created a generation of tracks that use short notes in strings in a way that is never possible to be reproduced by real players. Tracks are being written with an esthetic in mind that is not reflected in reality and it has become common practice to layer additional spiccato samples on top of the real recording to "match the demo better". This whole sample esthetic has in this case created a completely unrealistic expectation of short notes on string instruments.

If you closely observe such quick spiccato passages played by real orchestras, you will notice that even if they are marked ff they hardly ever surpass what would be a mezzoforte in the sample world. 

A few years ago, in my youthful naivety, I noticed this in one of my first scoring sessions and was absolutely sure that they were just not taking the dynamic markings that I wrote in serious so I requested them to play that spiccato passage even louder. What happened was that it ended up being only marginally louder but at the same time more strained sounding and considerably less precise. So I ended up reversing my request and questioning the for first time whether working with samples has spoiled my inner imagination of the real thing.

But unfortunately this abomination of Frankenstein spiccato passages became more and more prominent in scoring. If you deliver a sample based "epic" track to a mixing engineer or a dubbing stage nowadays, you will most likely get the request to isolate the string shorts on a different stem than the string longs so they can treat these differently in the mix and make them even larger in sound. And it seems like this is a self sustaining feedback loop with no end in sight with sample developers sampling excessivley large string line-ups to squeeze out more "epic" from the isolated spiccato samples.

Now my personal approach in music is to stay more on the realistic side so even in the loudest passages I hardly go beyond a strong mf sample for fast spiccato passages and would prefer to raise these in volume a bit rather than triggering a higher velocity layer on the samples if I need them to be a bit more present.

My personal wishlist for sample developers has a "performance spiccato" or "performance detaché" patch realtively high at the top. We can expect performance legatos with every new library that comes out where they sampled the transition from one note to the next note as well, but there is hardly any developer who tackled doing the same with short notes. Even though there technically is "nothing" in between these notes to sample, just the approach that the players will take to play two adjacent notes will make the sound different.

Now I know that VSL has already approached performance detachés which however lack (at least in my opinion) a lot on the sonic side so this item on my wishlist still remains intact.

As a bottom line, it is important to re-adjust your brain to "real" spiccato sounds and be conscious about the fact that "epic spiccatos" do not exist in the real and common orchestral practice. Of course when you have the resources to hire large line-ups and record several stems, you can create that effect but you will hardly have the luxury of doing such a thing. The most important thing is to develop a sensitivity for the differences between these two worlds as this will otherwise create unpleasant surprises once you record with real players.

By the way, this issue of sampling isolated short notes also happens on other instruments, however it is not so much common practice to write rapid staccatissimo melodies on for instance trumpets than it is on strings so that problem, while theoretically present, doesn't have as much influence on practical life in these sections.