Scoring Central

Full Version: Reverb Tails in Sample Libraries
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
I recently had a little adventure that I found pretty exciting and figured some might benefit from it as well.

We all use a variety of sample libraries together in our templates and so we all face the issue of "baked-in reverb." Libraries often feature wildly different amounts of recorded reverb on the samples. Our many conversations on reverb can attest to the many approaches and challenges in that area and how, particularly, it effects our own use of applied reverb.

Two of my main libraries are Project SAM "classics" (Orchestral Brass Classic and True Strike 1). Both are for the most part very wet. Perhaps because of the era in which they were released; I'm not entirely sure. Throughout the course of the few projects I've completed I've tried a variety of things, but I've noticed a steady trend toward drying up the samples as I move forward and learn more about reverb and space in a mix. It seems they sound great "out of the box" but as one's ear for VO develops one begins to notice certain things, call them "conflicts."

In the case of Orchestral Brass Classic I had recently begun to err on the side of shutting off the release trails (or sometimes, "reverb tails") on the sustain articulations and to browse the library, experimenting with the other patches for similar programming. I noticed that the release tails were really only present on the sustain samples and no where else (and not even all of the sustains). They were, in effect, a kind of "extra feature" that actually disrupted the continuity of the library over-all.

My "adventure" was in recently discovering that there was a bug I'd somehow only just caught in which many of the staccato and stac-variants did not dry up at all because the knob assigned for that did not work (needless to say I had this fixed by ProSAM post-haste). I had often had an issue with the roominess of some of the staccato articulations in OBC and suspected they were contributing to mud in the reverb of my tracks. It was rather exciting to discover I wasn't wrong! It's since led me to a revelation that there is a whole world of "baked-in" reverb on such samples as these, as well as one-shots and percussive samples that I really should be tweaking in order to better fit these samples to my own projects. I've advanced in my understanding of the relationship and differences between "stage ambience" and release / reverb tails. (This may be common sense to some here, but in my defense VO is complex and not all discoveries are immediately apparent). I've worked with samples before where it was obvious a little cutting back on the release was necessary, but in other cases it was almost as if my ears were perceiving the releases as part of the natural tone of the instrument. So there is another wrinkle on the brain for me.

Now, I'm not saying that releases are in any way a "bad" thing. I think it simply varies from library to library and are an important thing to consider in one's own mix, especially as pertains to the application of reverb.
I agree that release tails are problematic when mixing different libraries and I always turn them off or, if I'm mapping the samples myself, don't use them. Even if you don't get nearly as a realistic result from a simple envelope release, at least you have some fine control over it. You are also quite right in your observation that "perceiving the releases as part of the natural tone of the instrument". In a sense, they are. Too short a release time and the staccato note won't have time to breathe and develop and just becomes an unnatural blip. Orchestral instruments aren't recorded in an anechoic chamber for a reason, and that is because a dry room isn't the instruments' natural environment. Even libraries that one considers dry usually have some amount of ambience. You can't really take all of it away without it sounding odd. So while I agree that some libs with baked reverb are too wet and reining the release in is in many cases a necessity, you shouldn't overdo it either, cutting the note off before it has a chance to start decaying naturally.
I've often found a careful balance must be had with release tails/releases in general. I tend to actually like rather prominent release tails, but then again, I typically record a much closer, drier sound than most, even when recording in a concert hall. Good release tails implemented well, with good playing, can almost do as much for realism as true legato I have found. It really is something you notice is missing when you get a library that is either inconsistent with them or missing them entirely.

Here's an old test from VSCO 2's early days when I was doing some release tail tests on a tenorhorn patch-

Compare/contrast to a live recording (I think different mics though, but same space/instrument)-

For that reason, I am hesitant to suggest deleting/ignoring release tails altogether. Obviously if they are very long and wet, yes, probably more trouble than they are worth... but if you do remove them, make sure you give a good release length of minimum 150 ms. If you're working in SFZ, it seems to want longer numbers than Maize I have found (must use a curve rather than linear?), so I sometimes put releases on some instruments as high as 400-500 ms, but that is typically too much in Maize. It's sort of a 'by ear' thing, but proper release time makes a huge difference in believeability for every instrument. In general, I think you can over-compensate the release length a little if you plan on adding reverb, especially with instruments like piano which already are decaying at a certain rate.

I have heard in many cases 'in the wild' some orchestral patches that just have too short (e.g. 0-100 ms) or too long (e.g. 4/500+ ms) of a release time on the actual sustain sample. In both cases, it ends up synthy sounding. The VCSL SFZ patches make extensive use of this "poor man's release tail" technique, and blended with reverb, it's really almost indistinguishable to genuine release tails and can be transplanted to almost any space.

On the other hand, there are some instruments where release samples are absolutely necessary for a remotely realistic sound (e.g. harpsichord, drum rolls, many stringed instruments)-
Oh yes, I wasn't trying to suggest that I was on a quest to remove natural reverb entirely, I've just "leveled up" a bit in my understanding of how to listen for and control it. I completely agree that anechoic recordings are just odd. A good example of that can be found in the stock Kontakt brass, which apparently WERE attempts at completely dry recordings. They sound mostly lifeless and the trumpets sound almost comical. So obviously the idea of drying things up too much didn't really fly. In the case of OBC, some of the stac samples' tails were doubling up the tails of the Reverb I had dialed in myself which was basically just causing the decay to be louder. A little careful decay tweaking really helped reduce mud. For some reason I had missed that and was basically barking up the wrong tree for a while wondering if it was something to do with my reverb or EQ or something along those lines.