Ralf, higher sample rates for recording and processing do help with the aliasing effects caused by non-oversampled implementations of software synthesizers, plugins, and any nonlinear aspects of the signal chain, such as gain automation.
A non-bandlimited square wave synthesizer, for example, can sound noticeably different at 96 kHz sample rate than at 48 due to lower nonharmonic aliasing in the audible range. Particularly if played in upper octaves. That's not a very interesting synthesizer but it's a decent demonstration.
Compressors and other plugins that perform ring modulation/signal multiplication can also have this effect, though at lower magnitudes.
Aliasing effects can be avoided by oversampling internally, but if you start and end at 2x sampling rate you don't have to worry so much if every element of the signal chain has been audited for nonlinear operations.
For final output, I agree completely that 44.1 or 48 are fine. I personally have never blind ABXed 44.1 vs 48, but on pro and prosumer equipment I have used I have never been able to distinguish reliably between them on program material. I grew up in the CD age so I always think of 44.1 as the "real" final output and prefer to skip 48 completely so as to leave out a SRC step. I also work at 88.2 for recording to make the final SRC trivial. That's just a personal bias.
On Apr 2, 2013, at 8:44, Ralf Mardorf wrote:
> Peder I'm 46 years old and worked most of my live as audio engineer, so