[SPINOFF] Necessity of realism in sample libraries

[Crowe]

For many working composers, this is not a practical scenario.

If your gig is to create orchestral simulations, maybe you should try to….simulate an orchestra as best you can?

I get that it's not a practical scenario, which is why hunting for realism for realism's sake is a fool's errand. It's just not feasible when you realize that the time required for programming and recording all the minute variations that would be required for any basic level of true realism in a sample library would potentially only be outshone by the many, many terra-bytes such a library would then by necessity have to encompass.

I've seen the whinging about 200gb string libraries. Those would seem like a joke.

With what we have on offer, frankensteining a multitude of libraries into the type of cohesion that, even at surface level, looks like realism takes months for a single piece. We've seen it done. It's madness. Worth it maybe only to prove a point.

If that's the baseline we're working with, maybe it's our outlook that needs work, rather than our sample libraries.

EDIT: I'm sure Hans Zimmer has chimed in on this before with his own bespoke library. After which he was bashed into the ground by people who were dead certain he was a liar. This is why we can't have nice things.

 

[NYC Composer]

No, it is decidedly NOT a fool’s errand if you are being paid to create a reasonably realistic orchestral simulation. Achieving a consumer level realism IS the job in those situations!

 

[Crowe]

Achieving a consumer level realism IS the job in those situations!

Then there is no problem to begin with. Almost every sample library on the market worth its salt can do that.

 

[Dave Connor]

This conversation is hilarious. I'm going to be over here listening to Hiroyuki Sawano while I abuse my samples into the most unrealistic shapes I can manage them.

I've taken to live my life by the following adage:

If you require realism, go outside.

EDIT: In more applicable terms, that would mean: You should go work with musicians, not with samples.

I don't know anyone working in film on orchestral scores who isn't adamant about using the real thing. Sure we do mockups for the director but the orchestral sessions are what we live for. However, not all of us can afford to hire an orchestra for our own works and even with our mockups for live orchestra, we definitely want to filter out the awfulness that so often presents itself in samples. Not much harm in that even though it augers against nature in a way and may indeed be a bit of a fools errand in the end.

Then again there's this: Early on in the pandemic, I was working on the Netflix film, YES DAY with composer Mike Andrews, where we didn't know if our orchestral dates were going to happen or be cancelled. So we knew our orchestral mockups could very well be used in the film. We worked particularly hard from the start to bring them up to the best possible quality and sound. Sure enough, the orchestral recordings were cancelled and the mockups were used. That may be rare now but still, there are all kinds productions that can't afford a live orchestra (well that's what they say: there was a time that the lowest budget B movie had at least a $100,000 budget - before home computers etc.,) so there is indeed a valid reason for as much realism as possible for lower budget productions.

In the end, it's maddening to try to emulate the real thing and most of us far prefer the real thing. Someone like Hans Zimmer however, has managed to have his cake and eat it too. Which is to say, none of what he does is a compromise due to inferior quality or technical limitations but an execution of his exact intent: which he would not be able to do without sampling, synthesis and other processes in his arsenal of tools, all in service to his vision and soundscape.

 

[Casiquire]

I'm confused by this framing, as if it's a binary argument. It isn't. You do what's best for your music in context. Sometimes that's realism and sometimes it's not.

 

[biomuse]

I'm confused by this framing, as if it's a binary argument. It isn't. You do what's best for your music in context. Sometimes that's realism and sometimes it's not.

As far as I can tell the question seems to be about how important "realism" is. Some seem to be suggesting there's no intrinsic value to it at all with respect to certain musical goals.

At first glance I find that hard to disagree with. But my reservation stems from the fact that we revere the live performance so much in the first place. From there, I'd question whether it would ever be a bad thing to have the (making up a number here) thousands of degrees of freedom in producing bowed string sounds that a real string quartet has. Or even a significant fraction of them. Horses for courses, obviously. But the more horses, the more possible courses generally speaking.

 

[Eptesicus]

Yes, and Zimmer is often using samples to push beyond, even far beyond, what any real instrument could do. His use of samples is deeply musical, and that’s why we take it as sounding real even when it is highly fabricated, even when we can hear that it has been highly fabricated. The fact that they are samples ceases to matter because we are convinced by the music.

The answer to music sounding fake isn’t generally to make the samples sound more real, it’s to make them sound more musical, to find material that unlocks the musicality of the samples. Or you might say that in the case of someone like Zimmer that sampling is part of developing the musical material. He is recording and sampling to find and unlock musical potential in the material and so generates the samples that possess the character his music needs. In any case, realism is the effect of the musicality not the cause.

But realism in this sense has almost nothing to do with the way VI-C understands the term, as conformance of a sample library to what it emulates so that any musicality comes to seem the effect of the emulation, whereas in my view it is precisely the opposite.

Zimmer has indeed been one of the composers in the back of my mind as I’ve formulated thoughts about this.

But you are proposing two different use cases/scenarios.

If you are making samples of real instruments for a specific purpose or sound (IE NOT to just emulate a specific instrument) then obviously "realism" doesn't matter. You are specifically making a sample library to do something new or unique etc. I mean, there may be a reason you actually want a more synthetic sound, and that could be perfectly musical.

If I want a virtual violin section to replace a real violin section as I can't afford a real one, I want as much realism as possible and in that use case, the more realism, the more musicality. They go completely hand in hand.

To be honest I think you are massively over thinking all of this.

 

[Eptesicus]

I'm confused by this framing, as if it's a binary argument. It isn't. You do what's best for your music in context. Sometimes that's realism and sometimes it's not.

This.

 

[Dave Connor]

This.

I think that the fact that those who doing orchestral realizations regularly with samples and are constantly bucking up against the drawbacks and even sabotaging of their writing by the samples limitations have a need to discuss it. In fact, attaining realism with samples also dominates discussions of Notation programs and DAW programs, to say nothing of the fact that every discussion about sample libraries - new, old, or upcoming, is dominated by the hope that greater realism may be available, even if in a particular group of instruments or an advancement in technology generally in the sampling field.

 

[Elrik Settee]

Thank you everyone for a fascinating and serious and enlightening discussion.

My own comments were very tongue in cheek, I don't actually think that I have purchased an instant Zim thingy.

 

[TomislavEP]

To me, sample libraries are both a blessing and a curse rolled into one. Since I'm not classically trained and prefer working by recording and editing my performances rather than writing music, sample libraries are an indispensable asset. But if I had to choose and could afford to, I would still prefer using the real instruments and gear as much as possible.

Speaking of the so-called "realism" of sample libraries, I don't really care that much. IMO, if the music has quality, and depth, and evokes an emotion, the tools used in its creation become a relative thing. For example, whenever I listen to works by Vangelis, particularly his synth orchestrations, I'm completely mesmerized. And he never used sample libraries in his life that are otherwise vastly superior and incomparably more realistic sounding than his preferred sound sources. This is just one example.

I've stopped looking for that "holy grail" that some trained ear potentially couldn't distinguish from the "real thing" a long ago. This was never my goal in music anyway. I can agree that you should obey certain aesthetic guidelines when using a sample library. But achieving the "perfect realism" doesn't mean much to me, whatever it is.

 

[youngpokie]

But you are proposing two different use cases/scenarios.

Or many more than two - although of course it's very possible I'm misunderstanding what @jbuhler is saying. Even though the realism of samples is critical for me personally, I do not see his argument as blank rejection of that at all or as any kind of binary "either/or" choice.

To me his argument is about ultimately creating a different musical aesthetic (or several different ones) with orchestral samples, that no longer work exactly like a real orchestra or even the way orchestral writing conventions would usually work, but are nevertheless deeply musical and potentially even generative and/or transformative.

Today's samples are far more realistic than Roland TR-808 was for drummers. Drum machines were pretty bad at replicating real human drumming and the sound of acoustic drums. I don't know if drummers have a hot topic similar to "string legato" to endlessly argue about, but if this forum existed in 1980 it would have had a couple of threads on it. Either way, some musicians embraced those imperfect and unrealistic sounds for what they were - and went on to create a completely new aesthetic and eventually genres with their own rules.

The underlying process here - a new potential aesthetic that departs from the original "reality" to become a new genre all its own - is similar. So, while insisting on realism in order to better fit into existing conventions is a perfectly legitimate issue for a lot of people, others could also freely try their hand at finding new conventions.

 

[Jeremy Spencer]

Today's samples are far more realistic than Roland TR-808 was for drummers. Drum machines were pretty bad at replicating real human drumming and the sound of acoustic drums.

Except the TR-808 was never intended to sound like acoustic drums.

 

[I like music]

<MODERATOR NOTE - This was originally in the Black Friday Deals thread>

Let’s be real here. Even some of the supposed “worst” libraries by todays standards will still fool the average listener, who cares nothing about the technical details of a library, nor will they ever notice when a library has less velocity layers or unrealistic legato, etc. Most of them have no idea that those details are even a thing. They aren’t listening in that way. So the pianos we like or don’t like, which is all personal preference (meaning if somebody doesn’t like the “score” you give a library, it doesn’t mean your score was wrong), will almost all be good enough to our actual audience.

Those are things we nitpick here. They really don’t matter as much as we would like them to.

Brent

I think it is even more important to get the realism aspect right if your audience is the average listener!!!

I haven't read any of the many replies, but that's because I'm at work and I just want to answer what I consider to be a detail that you haven't given priority to, which is this:

The average listener may not consciously know or care, but they may well have internalised enough knowledge over the years (of listening to the actual thing, or at least a properly played real instrument) that something about a bad mockup will feel off to them, even if they don't consciously know it. In fact, in their case, it'll be even worse. They may have a bad reaction to the music and not realise that it is partly because the samples were bad, inappropriate, badly mocked up etc and they may mis-ascribe that to it being a bad song, bad composition etc.

Does a strings run done with a legato patch sound bad to us VI-controllers? Yes it does, but at least we know why. Will it sound off to a "non musician"? I reckon something won't feel right. But even worse, they won't know that it was a bad legato patch, and so they may mis-ascribe to the composer a fault that wasn't a fault in the composition, but in the production.

(Also, there's a good deal more going on in the heads of the average listener than we give them credit for. They feel music just as well as anyone else..)

 

[youngpokie]

Except the TR-808 was never intended to sound like acoustic drums.

I'm sure you know best, but according to Roland:

 

[koolkeys]

I think it is even more important to get the realism aspect right if your audience is the average listener!!!

I haven't read any of the many replies, but that's because I'm at work and I just want to answer what I consider to be a detail that you haven't given priority to, which is this:

The average listener may not consciously know or care, but they may well have internalised enough knowledge over the years (of listening to the actual thing, or at least a properly played real instrument) that something about a bad mockup will feel off to them, even if they don't consciously know it. In fact, in their case, it'll be even worse. They may have a bad reaction to the music and not realise that it is partly because the samples were bad, inappropriate, badly mocked up etc and they may mis-ascribe that to it being a bad song, bad composition etc.

Does a strings run done with a legato patch sound bad to us VI-controllers? Yes it does, but at least we know why. Will it sound off to a "non musician"? I reckon something won't feel right. But even worse, they won't know that it was a bad legato patch, and so they may mis-ascribe to the composer a fault that wasn't a fault in the composition, but in the production.

(Also, there's a good deal more going on in the heads of the average listener than we give them credit for. They feel music just as well as anyone else..)

I'm not disagreeing that realism is important. Since this post was a breakout of a conversation in another thread, it misses a ton of context. There was a conversation where somebody rated the EastWest pianos in their own personal rating "system", and somebody else came and disagreed and wondered how anyone could like those libraries, saying the pianos weren't as realistic because of this and that. A few more posts ensued referring to the need for certain technical "features" for a piano to be more realistic. Something that was mere opinion turned into something bigger.

I merely pointed out that as composers here on VI, we tend to nitpick things to a level that doesn't actually matter in the end. I have never suggested that we shouldn't try for realism, or strive to get better libraries for the sound we want.

I stand by all that. The conversations we have here surrounding things like, say, "realistic legato", are not about things that ACTUALLY matter to the average listener.

Case in point: The old EastWest Symphonic Orchestra was used on countless professional recordings. Guess how many legato samples it had? None.

How many workstation keyboards with tiny amounts of ROM were used for decades with pianos and other instruments, none of which had legato or fancy scripting? And how much do you think the average listener cared or noticed?

So my point is around the sometimes silly (though interesting and necessary in the context of this forum) conversations that happen here where we debate over little details. Nothing wrong with that. I do it too. But I wanted to add perspective that it doesn't really matter as much as we think it does.

Brent

 

[I like music]

I'm not disagreeing that realism is important. Since this post was a breakout of a conversation in another thread, it misses a ton of context. There was a conversation where somebody rated the EastWest pianos in their own personal rating "system", and somebody else came and disagreed and wondered how anyone could like those libraries, saying the pianos weren't as realistic because of this and that. A few more posts ensued referring to the need for certain technical "features" for a piano to be more realistic. Something that was mere opinion turned into something bigger.

I merely pointed out that as composers here on VI, we tend to nitpick things to a level that doesn't actually matter in the end. I have never suggested that we shouldn't try for realism, or strive to get better libraries for the sound we want.

I stand by all that. The conversations we have here surrounding things like, say, "realistic legato", are not about things that ACTUALLY matter to the average listener.

Case in point: The old EastWest Symphonic Orchestra was used on countless professional recordings. Guess how many legato samples it had? None.

How many workstation keyboards with tiny amounts of ROM were used for decades with pianos and other instruments, none of which had legato or fancy scripting? And how much do you think the average listener cared or noticed?

So my point is around the sometimes silly (though interesting and necessary in the context of this forum) conversations that happen here where we debate over little details. Nothing wrong with that. I do it too. But I wanted to add perspective that it doesn't really matter as much as we think it does.

Brent

Oh yeah. I see! Fair play.

I can't prove it but I do feel that non-musicians would still sense if the musicality was being appropriately communicated through a virtual instrument (or not being appropriately communicated). But maybe the effect is not nearly as pronounced as we think it'll be.