Sampling drums, when to make the cut
-
- KVRAF
- 1743 posts since 3 Dec, 2004
I'm working on a drum sampling project right now & I have a general question for the drum sampling gurus out there.
Upon looking at some commercial (and non commercial) samples of drums I notice that many people seem to leave a ridiculous amount of dead space after every sample.
Using my lil zoom too in Audition I then realized that this wasn't actually dead air, but very very quiet decay from the drums (<-96dB).
This strikes me as being a bit ridiculous, if more than 2/3rds of a sample (or even a third) is below this level, what is the actual point? Instead of a 100k wav, you have a 300k wave of mostly inaudible white noise? Instead of a 30 meg library, you have a 90 meg library where 60 meg is nearly inaudible except at extremely high volumes! Not to mention that you probably wont even hear this in a mix.
So am I crazy for thinking this way?
Upon looking at some commercial (and non commercial) samples of drums I notice that many people seem to leave a ridiculous amount of dead space after every sample.
Using my lil zoom too in Audition I then realized that this wasn't actually dead air, but very very quiet decay from the drums (<-96dB).
This strikes me as being a bit ridiculous, if more than 2/3rds of a sample (or even a third) is below this level, what is the actual point? Instead of a 100k wav, you have a 300k wave of mostly inaudible white noise? Instead of a 30 meg library, you have a 90 meg library where 60 meg is nearly inaudible except at extremely high volumes! Not to mention that you probably wont even hear this in a mix.
So am I crazy for thinking this way?
-
- KVRAF
- 10597 posts since 13 Jun, 2004 from Alberto Balsam
I cut it off where it goes under one pixel in my audio editor when zoomed all the way out & normalized (sorry if that doesnt make sense)
-
- KVRAF
- 10597 posts since 13 Jun, 2004 from Alberto Balsam
However leaving a bit more would be wiser as when you compress drums you can really bring the tails out
-
- KVRAF
- 1868 posts since 26 Oct, 2002 from San Francisco
I hope you work at a high resolution! Working at 640x480 would differ greatly to 1600x1200Chase wrote:I cut it off where it goes under one pixel in my audio editor when zoomed all the way out & normalized (sorry if that doesnt make sense)
-
- KVRAF
- 10597 posts since 13 Jun, 2004 from Alberto Balsam
I'd say it's fine. Here is a completely zoomed out Screenshot of a sample with the marker at the approximate point at which the waveform is quiet enough to fit under one pixel.jackson wrote:I hope you work at a high resolution! Working at 640x480 would differ greatly to 1600x1200Chase wrote:I cut it off where it goes under one pixel in my audio editor when zoomed all the way out & normalized (sorry if that doesnt make sense)
Here is a very zoomed in view of where the marker is in the first picture with the surrounding +/- peaks highlighted, peaking at about -70 db (quiet enough for me)
-
- KVRian
- 1001 posts since 6 Sep, 2005 from london
Well, there's no right answer, its down to what the library is meant to be conceptually.
If you are aiming to achieve a completely natrual, pure representation of the drum kit, then you might want to leave the tails intact. this is, after all , how the drum itself actually sounds.
Or , you might want to take an optimised approach and apply quicker fade outs to the tails, to reduce memory and polyphony, which, if done well, woudl sound perfectly fine.
To be honest, the real problem is with cymbals, which can ring for over a minute. Personally, on our drum library, we faded before that, and also set the polyphony of the cymbal groups to about 16, to stop cpu's going crazy. Applying temperance with the sampler this way is less destructive, and gives the user the choice.
If you are aiming to achieve a completely natrual, pure representation of the drum kit, then you might want to leave the tails intact. this is, after all , how the drum itself actually sounds.
Or , you might want to take an optimised approach and apply quicker fade outs to the tails, to reduce memory and polyphony, which, if done well, woudl sound perfectly fine.
To be honest, the real problem is with cymbals, which can ring for over a minute. Personally, on our drum library, we faded before that, and also set the polyphony of the cymbal groups to about 16, to stop cpu's going crazy. Applying temperance with the sampler this way is less destructive, and gives the user the choice.
-
- KVRAF
- Topic Starter
- 1743 posts since 3 Dec, 2004
Thanks for the suggestions, excellent point soniccouture regarding polyphony. I guess being (older) I tend to lean toward the optimized approach but at the same time I don't want to cut drum or cymbal tails too drastically.
Thanks for your pictures too Chase, I would tend to agree that -70db is more than quiet enough and my approach to finding a good cut point is very similiar to yours, although sometimes i'll just use AA's "delete silence" set around -80 db. Usually I'll apply a quick fade at the end as well to ensure no pops/clicks..
Thanks for your pictures too Chase, I would tend to agree that -70db is more than quiet enough and my approach to finding a good cut point is very similiar to yours, although sometimes i'll just use AA's "delete silence" set around -80 db. Usually I'll apply a quick fade at the end as well to ensure no pops/clicks..
-
- KVRAF
- 1868 posts since 26 Oct, 2002 from San Francisco
True. Hell, you write DnB, right? The dynamics of the tail of an already insanely processed snare drum are hardly the most pressing concern!Chase wrote:I'd say it's fine. Here is a completely zoomed out Screenshot of a sample with the marker at the approximate point at which the waveform is quiet enough to fit under one pixel.jackson wrote:I hope you work at a high resolution! Working at 640x480 would differ greatly to 1600x1200Chase wrote:I cut it off where it goes under one pixel in my audio editor when zoomed all the way out & normalized (sorry if that doesnt make sense)
Here is a very zoomed in view of where the marker is in the first picture with the surrounding +/- peaks highlighted, peaking at about -70 db (quiet enough for me)
If I'm being really fussy, I'll boost the volume like crazy, audition the tail and where the tail hits the noisefloor, cut it right there. Usually though, I'm more than happy to cut it a little short and add my own space (reverb, etc) in the track if I need a longer decay.
-
- KVRAF
- 10597 posts since 13 Jun, 2004 from Alberto Balsam
Exactly. Thats why i recommend leaving more space than my system for the sample tails for people that produce music that has more natural sounding drums, as I never have drums by themselves (DNB is a dog-eat-dog world when it comes to sampling - you have to cover up your shit!) so the rest of the track would never be silent for you to here that much detail in the decay of a sample.jackson wrote: True. Hell, you write DnB, right? The dynamics of the tail of an already insanely processed snare drum are hardly the most pressing concern!
-
Reverse Engineer Reverse Engineer https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=9129
- KVRAF
- 4968 posts since 23 Sep, 2003 from Glasgow
Where can we hear your music anyway chase?
-
- KVRAF
- 1868 posts since 26 Oct, 2002 from San Francisco
Speaking of dog-eat-dog, I'm heading home, gonna stop off for a hotdog on the way and then get to ripping a bunch of beats from some new house records I've been meaning to sampleChase wrote:Exactly. Thats why i recommend leaving more space than my system for the sample tails for people that produce music that has more natural sounding drums, as I never have drums by themselves (DNB is a dog-eat-dog world when it comes to sampling - you have to cover up your shit!) so the rest of the track would never be silent for you to here that much detail in the decay of a sample.jackson wrote: True. Hell, you write DnB, right? The dynamics of the tail of an already insanely processed snare drum are hardly the most pressing concern!
Mike
-
- KVRAF
- 10597 posts since 13 Jun, 2004 from Alberto Balsam
Do to being VERY, very busy I havent finished a track in ages, But here is some of my older stuff (4-8 months old, and therefore considered unfinished by meWhere can we hear your music anyway chase?
DnB:
1234
Non-DnB:
123456
Up-coming unfinished tracks:
123
Oi, now I feel as if I've Lady J'd up this thread
-
Reverse Engineer Reverse Engineer https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=9129
- KVRAF
- 4968 posts since 23 Sep, 2003 from Glasgow
Nice breaks chase. Battery was def the cleanest / best produced one imho. Saying that, sparkle was the track that appealed to me most.....I'm a tad jaded with the d&b scene though, so that was a nice repriseChase wrote:Do to being VERY, very busy I havent finished a track in ages, But here is some of my older stuff (4-8 months old, and therefore considered unfinished by meWhere can we hear your music anyway chase?).
DnB:
1234
Non-DnB:
123456
Up-coming unfinished tracks:
123
Oi, now I feel as if I've Lady J'd up this thread![]()
-
- KVRAF
- 10597 posts since 13 Jun, 2004 from Alberto Balsam
- KVRAF
- 1597 posts since 15 Jan, 2005 from Vales Of Glamorgan, South Wales, UK
Don't get me startedbirrbits wrote:This strikes me as being a bit ridiculous, if more than 2/3rds of a sample (or even a third) is below this level, what is the actual point? Instead of a 100k wav, you have a 300k wave of mostly inaudible white noise? Instead of a 30 meg library, you have a 90 meg library where 60 meg is nearly inaudible except at extremely high volumes! Not to mention that you probably wont even hear this in a mix.
Not at all.birrbits wrote:So am I crazy for thinking this way?
Chase and Soniccouture's advice is well given.
Steve
