[Yoozer]

Alright, here we go.

Setup
PC, Ableton Live 8, PC (Q6600, 4gb internal, Windows XP), a nice sample library, and an evening with time to spare.

Disclaimer
This test is by no means exhaustive or absolute, and I've mainly focused on the mono loop after the short stereo test because percussion samples are generally mono.

Method
All comparisons feature Ableton's audio track with the Inverter utility that flips the waveform. For the test, it shouldn't actually matter where I put it; I leave it on the audio track because that's constantly there, while I exchange the plugins. The limit is set at 2 tracks; the audio track and the plugin track. There's a Spectrum Analyzer on the Master buss to see if things happen my ear doesn't catch, and I'm monitoring on a pair of Adam A7s fed by my E-mu 1212m soundcard.

Ableton's Audio Track
First, a small test with Ableton itself. Two audio tracks, both the same stereo loop. One loop is checked with "warp", the other is not. The other has an inverter on it. When neither are changed in pitch, they null when the algorithm is set to "Beats", "Tones" or "Texture" - you'll start hearing something when it's set to "Complex" or "Complex Pro". In other words, if you're not stretching, don't use these algorithms as they can obviously change what you're hearing.

Hi-Q doesn't make a difference either if the sample isn't pitched up. Pitching both samples up to +3 semitones and leaving one on with Hi-Q results in similar artifacts as what I'm hearing when Complex / Complex Pro are left on; the loop sounds like turning down all the mids and lows.

Before using the samplers, I try to eliminate potential troublemakers; volume, polyphony, and of course velocity. Filters are bypassed.

Ableton's "Simpler"
Playing back both samples simultaneously results in an audible click at the front; most likely to avoid starting at a non-zero position, as well as a click at the end - most likely for the same purpose. The result is not null - any difference is around the -150 db, with the click at around -60 db.

Verdict: automagical adjustment is to blame for click. Perhaps not the most suitable choice for a percussion sampler.

Ableton's "Impulse"
The difference here is far more pronounced. The issue seems to be that I'm using a stereo loop.

Using a mono loop, soloing the single drum sound and setting the volume to around 3db results in peace and quiet - the highest peaks the analyzer shows are at -72 db, and perhaps I can get a bit more accurate when I type in a value. Switching "soft" on and off makes a difference again; like Simpler, it cuts off a bit of the attack to avoid clicks.

Verdict : summing engine and stereo samples are to blame for any differences in volume, but otherwise fine; nothing gets cut off at the start unless you check the "soft" button.

Vember Audio Short Circuit
This is one of my favorite samplers; it's a lot better in pitching up short material than Simpler is, in terms of aliasing. I'm testing the mono loop again, and I'm not getting any null. I like its footprint - far less of a hassle than Kontakt. On the other hand, it doesn't really come with any sounds, but that's not much of an issue.

The solution seems to be to temporarily set the pitch correction to -0.01 and then make it zero again when there's silence. Sadly enough, that moment in time is very hard to capture.

Verdict: I think the sample's getting a teensy eensy bit delayed tue to the filter/amp EGs - even if they're off, they are present in the structure of the sampler. Since it's a very small delay, the advantages of having the whole load of filters and clean transposition outweigh the problem of not starting exactly in time. Besides, it's not rocket science to compensate for that.

Native Instruments Battery
Finally, we're there! Setting the volume to zero and triggering the sample at HQI : Perfect (HQI is High Quality Interpolation) results in a perfect null. Absolutely nothing shows up at the spectrum analyzer; the Live audio track has Hi-Q set to off, but as the first example shows, that doesn't make a difference when the sample's not pitched.

When Battery is switched to HQI: Standard, something happens. I get a short click at the start and one at the end. While I can hear it, it doesn't show up on the spectrum analyzer, so it might be a single sample.

Verdict: Put Battery's HQI on Perfect and you've got no difference with a regular audio track, other than the volume which Battery helpfully puts at -6db so you don't clip rightaway.

Native Instruments Kontakt
HQI set to Standard gives a similar click to what Battery does. Also, I'd like to slap the NI team for not making it possible to type in values, since it's nearly impossible to set the volume correctly with such a tiny slider. It's either 0.1 or -0.5, but not both. Since the S770 works with an alpha dial, it's the more sensible approach: score 1:0 for S770 vs. Kontakt. However, using the audio track's volume slider I can get to null again; it's not easy, though.

Setting HQI to perfect oddly enough doesn't solve the issue with the click! Kontakt seems to have a different idea of what dB actually does, since the quietest result I can get is at -6.5 db instead of -6 db (Kontakt will only snap the volume slider of the instrument itself to -6db, not 0.

Verdict: You don't need the attack with the VSL library and the convolution reverb is absolutely awesome. There is however a difference that I can't solve and it seems to be purely volume related, and the click means that something's going on to soften the attack of the sample.

Final Verdict
There are several alternatives for audio tracks: Battery is one of them. Nulling proves its worth, despite bold claims and opinions to the contrary, since you're dealing with the digital domain. Use whatever you want to use; coloring is an excellent argument, the fact that a computer does word processing or internet on the side isn't.