Originally Posted by Aaron Giese
I'm not an expert on a/d conversion or anything, but I've recently been experimenting with 192 recording. I'm using an apogee ensemble, and i can hear a difference between 96k and 192k. Its definitely subtle, but to me 192 sounds more real.
Here's a thought: The Nyquist theorem says that you need a sample rate of at least double the highest frequency you want to reproduce right? And from what I understand the reason we use 44.1khz instead of just 40khz is because we need some room for the slope of a good cutoff filter to prevent aliasing. So that being said, in theory, what if we had an a/d converter that ran at 40khz sampling rate, and we wanted to record a sine wave at 20khz. The Nyquist theorem says that we should be able to do this accurately right? Well what if the sine wave and the a/d converter were running 90 degrees out of phase from each other? So the samples were being taken at zero crossings instead of at peaks and dips.
If that is possible, then I have to believe that 44.1khz isn't as perfect as everyone seems to think, and recording at 96k or 192k isn't just about recording higher frequencies, but also about capturing lower frequencies more accurately. In a perfect world, everything would be in phase with your a/d converter and transients would happen at the exact right point in time that your a/d converter is taking a sample. But in the real world, 44.1k gets pretty darn close, but doesn't always capture everything as perfectly as we think it does.
Like I said, I'm no expert. This is just a thought I've had and was wondering if its a valid argument for 192k or not. Can anyone else shed some light on this?
The theorem is correct. 44.1khz sample rates will capture and reproduce the human range of hearing perfectly. Saying "192khz sounds better 44.1khz" is a little shaky as a blanket statement, I think.
The flaw comes in the implementation. The sinc filter (the low pass filter applied upon converting the digital signal to analogue again) is idealistic in theory but in application it is more pragmatic. That is to say, the better the design of the converters and the sinc filter, the more negligible the difference between 44.1khz and higher sample rates becomes.
Originally Posted by jreatonit
You are right on the money. Imo(educated), if you can't hear it, then you don't know what your missing. Therefore 192k isn't useful.
IMO, if you can hear it... you don't know what you are missing...
...that would be converters with closer to idealistic implementation.