Hello Toontrack,
I wasn't sure what heading to use.
This is some I hope for you useful customer feedback. Please feel free to pass it on to your people internally, and I'm happy to discuss if this would be useful.
I have Superior Drummer, EZDrummer 2 and the Americana, Classic and Vintage Rock EZXs. I'm a hobbiest and I'm in my 50s - but please don't dismiss this part of your customer base!
I have two suggestions on 1. the way the SD sample engine works and 2. how the samples are recorded in the first place.
On 1., I tend to use my EZd libraries in the Superior engine. If, as is recommended by the manual, I have the humanise features all on, then when my midi gets played back, the samples are picked from the velocity layer of the midi note and the adjacent velocity layers. This means that for a typical snare back-beat at, say V=115, over 4 back-beats I might get a couple of different V=115 samples, but then I get one with a real crack to it from the "hard hits" layer above, and a softer (and duller) hit from the "gradient hits" layer below V=115. This results in snare hits with very different peak and RMS levels and each having very different timbres. Unprocessed, these don't sit in the mix properly. Real drummers may play inconsistently, but not this inconsistently. And - just to re-iterate - this is what the manual recommends i.e. leave all humanisation on all the time. If I experiment and turn off "Semi Seq" (which is the villain of the piece) I just get basically two samples from the gradient hits velocity layer of the midi note which sound "tick tock" alternately. Hmmmm.... So what I've ended up doing is two batches of audio bounces, one with ALL HUMANISATION switched OFF, and a second batch with ALL HUMANISATION ON except "Semi Seq". I use the first batch for the back-beat, because it produces a nice consistent beat without having to resort to compression / limiting / distortion. I use the second batch for snare fills (which btw also have various different midi velocities to add variety to the samples which SD/EZd selects). I do the same thing for kicks, where kick consistency is also important. The problem is - cutting and pasting the audio bounces is very laborious. So here's the suggestion for the next version of SD. Would it be possible to have a feature of playback which enables you to assign different humanisation settings to different groups of midi notes? This would enable me to have all my back-beats and main kicks (and indeed hi hats) totally NON-humanised and thus 100% consistent samples, but have my fills and anything intricate (e.g. semi-quaver hi hat parts) FULLY humanised? This would be a big improvement in IMHO.
Alternatively, it would be nice to have more - and "consistent-but-different" - gradient samples for each velocity layer. Which brings me to my next point.
2. Background: before buying SD and EZd, I attempted to generate my own drum samples, using a real drum kit and Logic's EXS24 sequencer. Pretty much all the issues which the SD/EZd sample engines attempt to address I came up against, and it was an interesting learning experience! But you guys are the pros and have the resources to record and package professional drum sample libraries: I obviously don't! HOWEVER, one thing I did learn is that there is a surprisingly big difference between the sound of e.g. a snare drum hit from a "cold start" and one that is hit when it's already ringing from the previous hit (if you see what I mean). The latter case is the way a real drummer sounds playing a real song: the start of the next hit is somehow "softened" by the end of the previous one. This is different to how I suspect drum sample libraries are recorded, which is to do a "cold start" hit, let it ring out to silence and then do another one - which I found produced a harder sound with a much more violent initial transient. Try it for yourselves and look at the waveform - and listen. You've get better equipment than I have but I suspect you'll find exactly the same thing. What I ended up doing (before I gave up on attempting to produce my own drum samples!) was to set a click around the tempo of the song, listen to this (quietly!) through headphones and then execute say a dozen back-beat snare hits at this tempo. I then sliced them up into individual hits and usually got about 6-8 decent ones that were similar enough (but not identical) which could be used in a song. I did the same exercise 3 times at the song tempo and slightly faster and slower (e.g. 115,120 and 125 bpm) so I didn't end up with samples that were slightly too short due to me being ahead of the beat). I'm sure you could do something similar and end up with drum samples which are less "bash-y" and - dare I say it! - more realistic. I think you already do this for hi hats (Seq Hits) so why not for drums too?
Hope this is useful.
Yours sincerely,
MarkImage may be NSFW.
Clik here to view.