Interestingly I bought a fireface 800 awhile back and experimented recording a four piece vocal ensemble. I tried recording using several different mic techniques at 44.1KHz,48kHz, 88.2KHz, 96KHz and 192KHz. When it came to finally mastering the material onto a CD I found that the best sound came out at 88.2KHz. I found the material I recorded at 192KHz to sound extremely harsh and abrasive.
Intuitivly I was thinking the old argument that when recording at higher sample rates your quantization noise gets spread over a larger spectrum.. thus when mastering down to CD quality audio you could low pass filter the audible spectrum and get rid of some of this quantization noise that was within the inaudible spectrum. This defineatly wasn't the case with 192KHz.... defineatly sounded like it was the case with 88.2.
In my own investigations of AD/DA technology I've been disappointed by nearly all sigma-delta converter designs and have set to working on an R2R non-oversampling ADC/DAC with 2 more bits of true measurement than the eccentric Dr. Altmann's creation ADC (who does a nice job of exploring AD/DA tech in layman's terms):
Lavry strongest argument against 192khz (IMO) is nearly all off the shelf converter designs halve the decimation rate as the sample rate doubles. [I think I'm stating this correctly, someone please correct me if I've forgotten the argument/logic]. One could conclude that there isn't any real gain to 176khz or 192khz unless it's done with a NOS true measurement ADC *OR* an approach like the Korg MR-1 is taken and the decimation and filtering can be performed in fake-time by software where latency is not an issue allowing many bits and excellent filter responses to be generated (thus not smearing transients and giving better reproduction).
Most of what I discuss above is prohibitive to the musician (Designing your own, Purchasing Altmanns @ 2000 euro's or the even more expensive true R2R 24bit master ADC by Lavry. Prism and Mytek are top designs but extremely expensive for a sigma-delta and out of the reach of most... And of course the Korg MR-1 isn't an interface and if it where it would have the exact same problems as any other sigma-delta system when you tried to use it with a PCM based software platform (Which is nearly every DAW, sans DSD editing platforms).
So, what's the consumer to do? After much research I've settled on:
A.) build my own NOS AD/DA (in process), when finished run at 96khz or 192khz (case/x/case)
B.) Use a Korg MR-1 as a master recorder for analog sessions (192khz derived from DSD using fake-time would appear to get around the halved decimation rate argument), me, I'm usually happy with 96khz.
C.) Use an echo Audiofire 12 for multi-tracking sessions with outboard gear/musicians
(the audiofire 8 and 12 both use the same crystal AD/DA as the Apogee Ensemble, which isn't that great in itself, BUT, for only 550$, it's a great value and is good enough for most music made today. +one can upgrade the Audiofire's analog path to get a decent sounding box, equal to or better than the Ensemble.
D.) Continue to use tape on specific projects although it's getting very expensive these days and sourcing tape in Tokyo is a nightmare.
This is 16 bits 44.1 kHz playback with various digital and filter techniqes, down to no filter at all.
The left shots show a step response, all right hand shots show a 8kHz sinewave playback.
What the diagrams tell us, is that there is a a choice:
EITHER you want a perfect sine, then you have ripple on every step input (its the filter that generates the ripple and its the ripple that makes the sine perfect). This is considered the correct technical way by many, and it is the way covered in all those digital audio educational books.
OR you want a perfect impulse (no filter) then your sine shows alias distortion (bottom pic).This is considered a better sounding approach by ... well some ;)
Now to the POINT:
With a higher sample-rate, lets say 88.2/96 or even 176,4/192 kHz it is - for the first time - possible to get a perfect impulse (by not using filters) and still have no alias distortion on the sine waves and these are the digital ingredients for fidelity recording and playback.
With respect to this insight, it is a pity, no... it is a shame, no... it is rather funny to see, that -at this time- no recording studio, no pro-audio manufacturer seems to make use of this potential.
Well, I think I have to take care that the following does not sound like a marketing phrase, but ...
... this also means and it is a definite fact, that nobody of the entire - what they call audio industry - has EVER listened to a TRUE 192kHz recording.
What they have listened to, is just the acid fallout that plumps out of their highly interpretational noise machines and this is no way to evaluate (or even enjoying) the potential of a 96 or even 192kHz sample-rate.