Digital audio requires anti-aliasing filters that (in the case of 44.1 and 48 kHz sample rates) roll off in the 20,000+ Hz frequency range.  It can be shown that these filters create rippling in the upper part of the audible spectrum, as it is impossible to design a filter that simply cuts off at a designated frequency without affecting  the area immediately below the cutoff frequency.  Higher sample rates allow the anti-aliasing filters to operate in a higher range and push the possible artifacts above the range of human hearing.  Others on the list may offer additional advantages, but this is one I have read a fair amount on.

Bruce Whisler

-----Original Message-----
From: Association for Recorded Sound Discussion List [mailto:[log in to unmask]] On Behalf Of Henry Borchers
Sent: Friday, August 17, 2012 10:37 AM
To: [log in to unmask]
Subject: [ARSCLIST] Bit rates over 48 kHz

I’ve been wondering what the real benefits of archiving music and spoken word at 96 kHz and above.  From what I understand, Nyquist–Shannon tells us that the highest frequency we can hear on digital recording is half of the sample rate, so 44.1 and 48 already covers the spectrum of our hearing range.  
I do know that high speed recordings at higher bit rate can be very useful for when you are trying to transfer large volumes of long running tape in a short time but I am curious what the rational and benefits are for transfer real time at 96 kHz. 

Henry Borchers
Broadcast Media Digitization and Curation Librarian University of Maryland