I am just playing around with a pop/rock track (completely based on MIDI and VSTs), and the drums were giving me some grey hair. The drums (Slates) always sounded strange and chaotic - like a drummer who is perfectly keeping the rythm, but not of control of the velocity. It turned out that the "round-robin" multisamples were simply too distict do facilitate good control of velocity levels.
I then switched to Battery, and it was fine. It is still not sounding too "mechanical" now, too.
If a good drummer plays four notes on the kick drum ("boom-boom-boom-boom" for some intro or bridge), he can do this perfectly constant. A "round-robin" drum sampler will give random fluctuations, that can sound extremly annoying. Additionally, e.g. if there should be a slightly different accent on the snare on "2" and "4", this effect can be completely ruined.
It is also impossible to correct for this in the MIDI file, as the use of the different samples is random and not predictable.
For sure, it is a much lesser problem as soon as any dynamics are squashed to death by compression and limiting.
Anyone else feeling annoyed by this effect?
Georg
Too much "round robin"?
-
- KVRist
- 397 posts since 1 Jul, 2011
There's often a switch that allows you to turn it off. Don't know about on Slate Drums.
I often find round robbin, when it's on a predictable cycle, becomes not much better than the machine gun effect. For example, when the third sample in a cycle has a slight delay at the start of a note, it leads to a lack of smooth timing when trying to play triplets. When repeating the same note say a dozen times, that slight delay turns into a quirk in the rhythmic feel of the playing, so that the repeating cycle becomes noticeable and unnatural sounding.
It's funny that you pointed out that a drummer could play the kick to sound the same time after time. I was just listening to a Sigur Ros song that has toms being played with mallets, and was thinking that it would be difficult to replicate that sound with samples as each hit in the recording sounded slightly different. Even though the hits are separated by a gap in time (so would not be affected by the machine gun effect), the reuse of the exact same sound would still be noticeable and still sound artificial.
I guess I have mixed feelings about it overall.
I often find round robbin, when it's on a predictable cycle, becomes not much better than the machine gun effect. For example, when the third sample in a cycle has a slight delay at the start of a note, it leads to a lack of smooth timing when trying to play triplets. When repeating the same note say a dozen times, that slight delay turns into a quirk in the rhythmic feel of the playing, so that the repeating cycle becomes noticeable and unnatural sounding.
It's funny that you pointed out that a drummer could play the kick to sound the same time after time. I was just listening to a Sigur Ros song that has toms being played with mallets, and was thinking that it would be difficult to replicate that sound with samples as each hit in the recording sounded slightly different. Even though the hits are separated by a gap in time (so would not be affected by the machine gun effect), the reuse of the exact same sound would still be noticeable and still sound artificial.
I guess I have mixed feelings about it overall.
-
- KVRian
- 628 posts since 18 May, 2010
Round-robin should of course imply that the samples general characteristics, such as level, attack, etc., match. If the samples differ too much, that's just a bad implementation.
Some libs have a key switch or CC which lets you reset the RR queue, so you can get a more consistent performance.
If you've got access to the sample files themselves, you could use an editor to increase the level of the offending sample(s) to match the rest (don't forget to save backup copies).
Some libs have a key switch or CC which lets you reset the RR queue, so you can get a more consistent performance.
If you've got access to the sample files themselves, you could use an editor to increase the level of the offending sample(s) to match the rest (don't forget to save backup copies).
-
- KVRian
- 616 posts since 8 May, 2012 from Sydney, Australia
I've had problems with round robin myself. For example, in a string library I have, sometimes I'll be trying to create a staccato section and just a couple of notes, because of round robin being on, sound like they are played incorrectly and make it sound out of time. So I often turn it off and just make sure the velocity levels have a bit of variation. It's a good thing when implemented correctly, but if it's a bit off, it can ruin what you're doing. I'd rather machine gun than bad notes. I'm sure some libraries do it very well though. I'm not sure if Superior Drummer has round robin in this way, but I've always found it to humanize very well.
-
- KVRAF
- 4314 posts since 31 Oct, 2004
Isn't the velocity with multi-sampled velocity layers enough to make realistic sounding drum sequences?
-
- KVRian
- 921 posts since 14 May, 2010 from Atlanta, GA
It usually is, SS, until you do snare or tom rolls that have the same velocity, then the single sample shows itself pretty clearly. I too have had some round robin problems with some smaller dev stuff, and some let you turn it off, (Forgotten Keys), while others don't (Wavesfactory).
It's REALLY frustrating to have something that's supposed to make the product more realistic make more work for you, or even have to use a different piece of software to get the job done. I find that velocity switching is a more reliable way to control articulations without key switching, and still avoid the machine gun effect by editing the velocities of a few notes.
KVR/eSoundz: Xenobt
It's REALLY frustrating to have something that's supposed to make the product more realistic make more work for you, or even have to use a different piece of software to get the job done. I find that velocity switching is a more reliable way to control articulations without key switching, and still avoid the machine gun effect by editing the velocities of a few notes.
KVR/eSoundz: Xenobt
-
- KVRAF
- 1735 posts since 28 Dec, 2007
it depends on how many samples over how many velocities and if you program beats with plenty of small velocity changes to use the different samples...SampleScience wrote:Isn't the velocity with multi-sampled velocity layers enough to make realistic sounding drum sequences?
-
- KVRist
- Topic Starter
- 120 posts since 13 Sep, 2007 from Austria
Thanks for the feedback, guys!
I for my self would always prefer an increase in verlocity layers over additional round-robin.
For me, the best solution would be a sample library with many velocity steps (at least 16, maybe 32) combined with the ability of the sampleplayer to apply random (and very, very minor) alterations of pitch, filtering, and envelope shape. This could be steplessly adjusted and would "humanize" the sound without being too obvious.
BTW, there is also a difference between hitting a percussion instrument on different areas (or with a different technique) to intentionally change expression, or to do so simply because someone is not in control of the sticks - and the typical, random RR usually covers the latter situation.
Georg
I for my self would always prefer an increase in verlocity layers over additional round-robin.
For me, the best solution would be a sample library with many velocity steps (at least 16, maybe 32) combined with the ability of the sampleplayer to apply random (and very, very minor) alterations of pitch, filtering, and envelope shape. This could be steplessly adjusted and would "humanize" the sound without being too obvious.
BTW, there is also a difference between hitting a percussion instrument on different areas (or with a different technique) to intentionally change expression, or to do so simply because someone is not in control of the sticks - and the typical, random RR usually covers the latter situation.
Georg
-
- KVRian
- 1181 posts since 6 Jun, 2002 from Southern Germany
round-robin is principally a good Thing, but I agree that it is sometimes exaggerated - especially in string libraries. when the variation is too obvious in timing and timbre. I think this all has to be programmed extremely carefully so that the unity of the samples does not break apart in a Musical line.
greetings
wolf
greetings
wolf
-
- KVRAF
- 2448 posts since 12 Sep, 2004
The RRon the addictive drums acoustic snares has always driven me nuts... Every so often I'll get a "one of these hits is not like the other" effect... Like all of a sudden a bit too much rimshot or whatever... I would never play something so obviously different just for the sake of "variation"... Especially on 2 & 4 rock/pop/etc. stuff, which more than any other type of playing should be very very consistent and without unnecessary complexity. I find the best tool for humanizing programmed drums is a good wealth of velocity layers and making subtle velocity and timing tweaks by hand as well as throwing in some articulation at natural points (a rimshot, flam, ruff, whatever)... But yeah, too much RR on percussive instruments can be a problem...
You need to limit that rez, bro.
-
- KVRist
- 175 posts since 3 Feb, 2005
In ideal sample library there is always plenty of velocity layers AND round robin, but it is essential that those round robins sound "almost" the same with each other. It is maddening when developers combine completely different type samples with RR. That can mess the whole rhythm.
With seemingly simple stuff like glockenspiels and such RR really makes big difference, IMHO.
With seemingly simple stuff like glockenspiels and such RR really makes big difference, IMHO.
-
- KVRAF
- 3220 posts since 4 Jan, 2005
I like round robin , without it on acoustic drums just sounds kind of robotic, especially on snare and hit hat , as well as other cymbals rides etc etc , I could almost do without round robin on kick and toms at times .