EXS 24 Logic Sampler Users Group

Herbert Bolandf wrote:

>
> Why do sampled instruments sound artificial in general? The ear
> subconsciously detects that something is wrong when individual notes
> sound 100% identical, which is impossible in nature. Unfortunately this
> still is the case (although to a lesser extent) with a lot of layering
> and samples per note, or 24 bits 192 kHz recordings. Maybe they should
> invent a mechanism that alters the attack/transients randomly on every
> note.


This is a limitation of the velocity switching model which EXS24 and its
competitors employ.

Take a look at Drumkit From Hell Superior - just released by Toontrack of
Sweden. It has 35 GB of samples on 9 DVDs and its own very sophisticated sample
engine. To get around this repetition of attacks problem it works like this:
for a give drum hit (eg stick on snare): rather than a simple velocity switched
set of 16 or 32 samples  it divides the velocity range into three parts -- 
Soft, Gradient and Hard. 

A whole set of say 10 samples are recorded for each of these groups. When you
play in the soft range - say a press roll and you have dozens of low velocity
hits in rapid succession , there is a randomising function which ensures that
the same sample is never repeated twice in succession and over the course of
the roll the set of ten soft samples will be cycled through at random. Same
with the hard hits. With a smart enough engine and enough samples and disk
space the repeated sample problem can be banished.

I think you still end up with the problem that 30 hits in a row on a ringing
ride cymbal is not quite the same as 30 ride samples overlapping each other.
This is related to some degree to the summing of piano resonances that has been
spoken about in this thread.

I suspect that the evolution of samplers has a long way to go yet.

Regards,
Murray