What do you want to see in a Kontakt drumkit sample library? (acoustic and electronic)

[Gonga]

I've been looking for a really high-quality set of drums that have a patch for each drum, and each drum is pitch-mapped to the keyboard so you can play melodies on them. This would include percussion such as cymbals and cowbells also. Basically I'd like to be able to play each drum as a totally separate instrument rather than a set.

But then, I might be the only one interested in something different like that.

This suggestion is more brilliant than you realize. I have my own ideas why this would be useful. Tell me, what sound are you going for? What genre of music? Tell me more.

I have Chromaphone, which is fantastic, but a modeled approach. Actually, I like instruments that sound different, so anything from a drum to a cymbal to any other sort of percussion, especially pitched, that sounds different is perfect for me. I'm not sure a sampled library can better Chromophone except for two things I know of. One, samples can capture complexities and nuances of sound that are tough or impossible to model. Second, pitch bends. It's tough to set up Chromophone for pitch bends. I like to pitch bend even percussion (not just pitched instruments but deep drums, shakers, anything). And I like to have separate control over the upward and downward bends (usually 12-24 semitones downward and 3 semitones up). Chromophone can't do that.

[Doug1978]

Hello guys,

I am an upcoming sample library developer interested in your opinions. What is wrong with drumkit sample libraries, what is missing, what do you want? Interface-wise, sample-wise, etc.

Edit: Oh also important, price-wise.

Hi, many thanks for agreeing to do this Architeuthis.

TBH, I have many drum libraries.

But I don't yet have one of the sound of farm animals. I love the sounds of nature and I imagine some wonderful 'baas' and 'oinks' would be very popular with other people as well.

Failing that, a drum sample library of old people muttering would be cool.

Cheers

[pljones]

The most important features by which I instantly value or disqualify drums, be they for Kontakt or not, are these:

All of those but make sure the hi-hat really, really works. 5 degrees of openness is no good. Nine is more like it. Make the pedal and foot splash authentic. If you need less than 2Gb for the hi-hat, you did it wrong.

Good luck, it's likely to nearly kill you getting it right . (And if the recording doesn't kill you, the drummer just might .) (No mis-hits except as extras, please!)

Oh, and I'd pay £125 ish for a really good kit.

[Architeuthis]

The most important features by which I instantly value or disqualify drums, be they for Kontakt or not, are these:

All of those but make sure the hi-hat really, really works. 5 degrees of openness is no good. Nine is more like it. Make the pedal and foot splash authentic. If you need less than 2Gb for the hi-hat, you did it wrong.

Good luck, it's likely to nearly kill you getting it right . (And if the recording doesn't kill you, the drummer just might .) (No mis-hits except as extras, please!)

Oh, and I'd pay £125 ish for a really good kit.

I hope to cater to electronic artists and those who don't have a drum-purist mindset, and samples that don't kill your computer. With that being said, your price point is way too high, there are many drum kits already available that one must give an arm and a leg for. That being said... I doubt any drummer could articulate 10 stages of hi hat openness. That doesn't mean it can't be sampled, but I wonder if fully-sampling 5 opennesses, and then sampling inbetween openness just to serve as openness-transition samples would be smarter. Another goal I have, which 10 oppennesses conflict with, is ease of use. I don't want the end user to have so many articulations and decisions that could have been just one humanized-easy-to-use-sounds-real articulation. You tell a drummer to hit the snare, do you also tell them where to hit the snare? A combination of artificial-intelligence scripting and sampling should take care of that.

BFD is a much better interface for drums than Kontakt.

Are you saying individuals like me have a potential to create products for BFD? That would probably cost money and my creative freedom. Yes Kontakt is not perfect, but it does have some advantages.

sorry for cheap marketing

Nope, feel free.

[jancivil]

there is one very essential thing missing, perhaps missing completely. at any rate I'm missing it and I have a lot to work with. a single bounce of the stick; not a buzz, a single bounce, and as left/right placed on the drum. so one can construct rudimentary drummer in midi. flamadiddles and all of this would be convenient, but I'm fine with making things happen. and you could extend it, three bounces up to whatever rep constitutes the 'buzz' articulation. two single strokes is not the same action off the head as a bounce, you can fake it with velocity but it isn' the same sound.

I'm not sure I know exactly what you're talking about. I know that there's a difference between hitting the snare with the intention to leave the stick on the head and hitting the snare with the intention to make a clean bounce, but I thought the drummer usually goes for clean single bounces.

anyone with proper technique is 'clean'. what I'm saying is that of all the BFD2 libraries I have, there is 'hit', 'flam', 'sidestick', 'buzz', 'rimshot'. None of these is a real basis for a measured 'double-stroke' roll.
notice at ~2:54 the roll on the floor tom. actually notice at the beginning, the cymbals work. that particular effect is going to be so much easier to do with the actual bounce vs 'hit'; which here he shows how to get more bounce off of less responsive drums than the snare. the floor toms particularly illustrate the sound of a double-stroke roll. this may be more ambitious than you want to go, but in terms of what's missing out there, that is number one for me.

What I have done with my upcoming products is included ghost notes and really quiet notes that do sound like clean single bounces where the hit is light enough and clean enough to the point you no longer hear the snare sizzle and it almost sounds like the "snare" part of the snare was removed.. Maybe this is what you're talking about?

Edit: But if you have specific articulation ideas, I'm all for it as well.

it might work, I have to reserve judgment on it just out of some words.

additionally what say the BFD libraries omit is scrapes and brushing of cymbals. the VSL percussion has a couple of very useful such articulations but they don't go all that far. That said a couple of things goes a long way. I think I don't agree with nine open hats articulations. the amount of velocity on thesee is itself going to force the closure attitude somewhat. However the really useful hats I go for in BFD2 are these 3 GB jobs.

since this is going to be inexpensive, I would go with boutique kind of items and innovate rather than cover ground the BFD2 libs do with the basics.

[Architeuthis]

a single bounce of the stick; not a buzz, a single bounce, and as left/right placed on the drum. so one can construct rudimentary drummer in midi. flamadiddles and all of this would be convenient, but I'm fine with making things happen. and you could extend it, three bounces up to whatever rep constitutes the 'buzz' articulation. two single strokes is not the same action off the head as a bounce, you can fake it with velocity but it isn' the same sound.

Ok, so if you use a standard snare hit articulation at different velocities close together, this is essentially a double-stroke, but the samples don't sound realistic enough because it's not actually a double-stroke? And this is what you want solved?

So, would this solve your problem?: Record double-strokes at different tempos which kontakt then automatically chooses the right-tempoed double stroke or flamadiddle.

since this is going to be inexpensive, I would go with boutique kind of items and innovate rather than cover ground the BFD2 libs do with the basics.

Indeed!

additionally what say the BFD libraries omit is scrapes and brushing of cymbals.

You'll have to tell me more!

[spirit]

Playability is one of my top concerns. I hate loops, but just having the sounds and nothing else can sometimes be daunting. The "uberpeggiator" in SoundIron's Apocalpse drums is fantastic. A fast way to get rolls and basic patterns happening.

[Architeuthis]

Playability is one of my top concerns. I hate loops, but just having the sounds and nothing else can sometimes be daunting. The "uberpeggiator" in SoundIron's Apocalpse drums is fantastic. A fast way to get rolls and basic patterns happening.

Oh man, yet another awesome suggestion and I've even been thinking about this. I would love to give the end user the power to customize midi-triggered sequences for 8th-note or 16th note-beats or other simple patterns, but this requires a host of gui planning and scripting, I'm not there yet. However, I plan on shipping my product with midi sequences to fill in this need for now.

[Shy]

The reason a real double stroke may sound quite different is that by hitting the drum in two places, it goes on to resonate in a different way than how it would resonate with one stroke. The "clashing" frequencies form a different character. Blending two different recordings of single strokes can't give a similar sound. I'd compare it to trying to make a single oscillator in a synthesizer sound different by blending it with another oscillator tuned differently instead of applying wave shaping of sorts to it, modulation of sorts, syncing to other oscillator/s of sorts, delay networks of sorts, etc. The result of two blended waves means you don't have a real independent source, you just have an unrelated blend that won't sound right and which you'll never be able to treat and build upon as with a single source.

I think this is all VERY problematic, same as with recording a hi-hat. You know that hi-hat cymbals are the most versatile part of a kit, and the amount of very different sounds you can get from it is huge. There is NO way to record hi-hat samples in such a way that they would be anywhere near as playable, versatile, and just plain "real" sounding as the real hi-hat. Even if you make it solely as a complex "programmable" library, still it would not be anywhere near, it would be horribly complex to handle, and most of all, dealing with the recording and midi note writing/sequencing of such a thing would be anything but a "musical process", it would be a nightmare.

The same goes for a snare drum. When you start to take stuff like multiple strokes into account, it quickly gets way too variable and in order to make something usable out of it you'd need to be far more innovative than what I could think of. I personally can't see a practical way to get a proper level of playability if anything more than fairly "basic" recordings is used in a drum sample library. It very quickly becomes a nightmare to handle if you attempt something that is truly "realistic". I've tried it myself in various funky methods, since I have a huge interest in getting real realism in a virtual environment, but my conclusion is that the only way to get there is with physical modeling, and anyone who really knows how drums and cymbals sound and react, would know that maybe some emulation mechanisms can go quite far, but the actual sound, there's a very, very, very long way to go until someone gets there..

I'd say as far as drum sample libraries go, first things first: at least, get the basics right like proper recording / sound, and enough versatility. Hardly anyone has done even that, so I wouldn't get ahead of myself if I was aiming at making a serious product.

[pljones]

I play my sampled kits live. I'll just say that I hate multi-hit samples - they never sound right played as the timing is off; I guess if you're typing the notes into a sequencer, it's okay. For hi-hat, yes, the continuous controller footpedal on my set up lets me access all 128 different positions and I can get accurately about 10. But I can feel the difference between ns_kit7 and every other library I've used, none of which can compete in terms of real playability. It's not whether I can always get exactly the position - it's whether the library supports a natural sounding hi-hat. If you're after a standard for live playing acoustic, I still put ns_kit7 forward every time. (Noting I've not used BFD or DFH sets at all.)

Most acoustic drum libraries aren't particularly useful...

Native Instruments Abbey Road kits are way over-priced. I still bought 60s Drummer. Don't use it much but it's handy now and then. For a simplistically sampled kit, the Analogue Drums libraries are a better price comparison point.

[Architeuthis]

...multi-hit samples - they never sound right played as the timing is off

Indeed.

[strayboom]

i have questions in my mind about "realism" and "live acoustic drums".
what if you don't need extra articulations? what if you only need pure drums? because you only want to do a simple drumbeat as a base for your song?
do you want to load an extensive library just because of pure dumm-tsaa-shhhh?
or do you need extra articulations because you want to imitate the übermegabrutal drummer?

to be honest, i found a lot of drum libraries sound-wise good but a bit too much. there are tons of extra features i barely use. also you (mostly) need to load the full set, but what if you need only the snare?

also, i'm not sure if the "classic" microphone set works for all possible music. i mean, sometimes i don't need the overheads, sometimes i don't need the toms, or i don't need to hear snare wires in the overheads, etc.

obviously, there are very good drum libraries available, feature and soundwise too.
maybe it's time to change thinking direction and create some different libraries, to offer unique and different sounds

that's why i did my drum sound library, recorded individual drumkit pieces and they are available to load separately, use microphones levels, panning and routing "as you wish".

Architeuthis, the question is, do you want to create the überdrumkit library, or do you want to create a unique set of sounds?
both can fill a gap and can be successful

[Architeuthis]

Architeuthis, the question is, do you want to create the überdrumkit library, or do you want to create a unique set of sounds?
both can fill a gap and can be successful

My goal is to cater to electronic (breakbeat, drumNbass, dubstep, trance, dance, techno) composers. So yes, the sound will be different. I will be including synth percussion and encourage the user to layer this with the acoustic percussion (which will be easy with fully custom key assignment), with a plethora of sound shaping possibilities. It will be modular (you may load up a snare, use whatever mics you want or don't want, with full-fledged sound shaping and mixing options per instrument and per mic per mic.) I will sample kit-at-once for some products, instrument-at-once for others, and a combination of the two techniques (stationary close mics, but moving overhead mics around, for example) I'll be doing every strange thing I can think of, even setting up 2 or 3 snares surrounding a kick just to make the kick it super-sizzly. Oh, here are my priorities in order of importance:

1. Usability
2. Sound quality
3. Customer support (taking suggestions, allow the customer to have control over products released, etc.)
4. Realism
5. YouTube videos (video-manuals, product demos)
6. GUI, art, visuals
7. Marketing, Sales, Business, (*throw's up*) etc.

[jancivil]

a single bounce of the stick; not a buzz, a single bounce, and as left/right placed on the drum. so one can construct rudimentary drummer in midi. flamadiddles and all of this would be convenient, but I'm fine with making things happen. and you could extend it, three bounces up to whatever rep constitutes the 'buzz' articulation. two single strokes is not the same action off the head as a bounce, you can fake it with velocity but it isn' the same sound.

Ok, so if you use a standard snare hit articulation at different velocities close together, this is essentially a double-stroke, but the samples don't sound realistic enough because it's not actually a double-stroke? And this is what you want solved?

So, would this solve your problem?: Record double-strokes at different tempos which kontakt then automatically chooses the right-tempoed double stroke or flamadiddle.

well it's the tempo thing which discourages this approach I believe, so yes. I have no idea what that entails for the developer. As to the basic question, a hit and a bounce is not a single hit and another single hit; it sounds different because it is different.

[jancivil]

It's not whether I can always get exactly the position - it's whether the library supports a natural sounding hi-hat.