[sdiy] harmonics & vibrato - drums

[Richard Wentk]

At 06:15 20/12/2002 +00-08, Scott Gravenhorst wrote:
>Ok, this makes more sense.  But the question remains, is there a random
>factor or not?  If not, then why does doing a *slight* randomization
>(weighted in my case) sound better (um more "human") than a quantized track?

Getting a touch empirical here - I'm wondering if you'd be able to tell 
randomised and non-randomised tracks apart in a blind test.

Not to be picky about your ears in particular, but I've certainly sat there 
tweaking things in a sequencer, thinking to myself the result sounds 
better, and then realising that for one reason or another my changes 
haven't been making any real difference at all. Ugh. :)

The other problem is MIDI's timing resolution which includes a certain 
amount of slop anyway, especially in places where a lot of events are 
stacked up. And there's also another source of slop at the other end (if 
you're using a hardware sampler or synth) where the processor takes a while 
to work out which notes it's supposed to be playing, and how hard. So MIDI 
is kind of randomised a little whatever you do - which is allegedly another 
reason people prefer drum machines. Whether you'd hear it as such depends 
on how the spiky the attack of the various sounds is, which ones will take 
priority in your ears, how good your ears are, and probably a whole bunch 
of other psychoacoustic voodoo.

Anyway, the point here is that 'better' may really mean the instruments 
involved become clearer because you're hearing timbral variation in the mix 
because no two hits are identical, not because you're hearing random 
variations in the timing.

Dunno if that's true or not, but it's something to play with.

>If there is no true random factor in human playing, then there seems to be
>a random-like factor, since the tracks do sound better with randomization.

Do you want to email me some before and after MIDI files and I'll have a 
play here to see if I can tell them apart without looking, and which one I 
prefer? Maybe other people might want to have a go too. Could be a good 
quasi-objective kind of test. Obviously it's going to be hardware 
dependent, but if you use some GM sounds for the drums the results should 
sound kinda sorta comparable.

>  Again, this is not the same as moving the strike before or after the beat
>for feel, I do this as well.  I will also move an entire track forward or
>backward in time by a constant amount for placement in the field.

I agree both of these make a difference. But I've noticed that some of this 
is because the perceived attack point on a sound like a bass may be quite a 
bit behind where the sound really starts. There's quite a bit of 
flexibility in placing sounds with attacks of say 20ms to 50ms, because the 
perceived attack actually depends on the context.

I remember Fairlight doing some research in musicalising sequence playback 
in an intelligent way before they went under. I don't think anything ever 
came of it, but I know they were definitely thinking along those lines.

Also the MIDI performance in Sibelius has a moderately smart performance AI 
feature which adds algorithmic feel to MIDI performance. It's pretty good - 
much more musical than straight playback.

Richard