SPEC Kit 329: Managing Born-Digital Special Collections and Archival Materials · 147
RUcore. Sound Objects
Audio/Video Standards Working Group: RU-CORE and NJDH Standards Analysis for Audio Objects
I. Beard, I. Bogus, N. Gonzaga, B. Nahory, R. Sandler 1 Last update: 8/9/2010
Recommended minimum requirements
for preservation sampling of audio
This document will set forth two standard requirements for audio. One will establish a minimum and
recommended sampling rate – the quality level at which the audio is digitized – for the digital audio masters and
presentation copies. The second standard will recommend specific file formats for the preservation master and
derivatives, for implementation into the Workflow Management System (WMS).
Although the standards will be different, the philosophy behind preservation and presentation will be
same as for all other object types. It will be mandatory to archive an uncompressed archival master, to ensure
an object of the highest quality is preserved. Additionally, a small but diverse number of presentation copies
will be archived as well. These presentation copies are to be stored and accessible in formats that the end user
will find easy to play back, and will be “low-bandwidth friendly” whenever possible, allowing users with
slower internet connections to have access to these objects as well.
Sampling and Digitization Rationale
As with all other objects, obtaining a high quality sample of the original for preservation in RU-CORE
will assure the best chance of long term preservation without having to go back to the original source for a
resample in the future. This will also allow us to ensure that the presentation copies provide a comparatively
high fidelity that sacrifices little in quality. In the digital realm, audio is represented by a digital sampling at a
set frequency, to obtain a granular but reasonably accurate representation of the analog original. Sampling is
the process of converting a signal (e.g., a function of continuous time or space) into a numeric sequence (a
function of discrete time or space). The higher the sampling rate – it is assumed – the more accurate the digital
representation will be.
For audio, there has been a wide practice of following the Nyquist-Shannon Sampling Theorem, a
doctrine which is used to assert that 44.1kHz is an acceptable minimum sampling rate for all audio. This belief
is based on the established fact that most human ears perceive sound up to an upper frequency threshold of
20,000Hz, and sampling must occur at twice the upper limit to achieve an acceptable digital copy.
Consequently, a number of digital recordings, including CDs, adhere to this standard sampling rate (thus the
term “CD Quality” is attributed to this sampling rate).
This 44.1kHz sampling rate is not without its detractors. Over time, audiophiles have consistently
complained that they perceive a loss of fidelity when analog recordings are digital remastered to CD Audio.
While some audio experts have insisted that these complaints are based on purely psychological factors, there is
some support for a need for a higher sampling rate. There are inherent risks in losing quality to the sampling
process, causing a degradation that is not accounted for in Nyquist. However, a higher sampling rate may be
able to compensate for these sampling losses.
As a result, the standard set forth accounts for the CD-Audio minimum sampling rate and accepts it as a
minimum, while recommending a higher level whenever the opportunity to sample at a better rate presents