Studying the effects of different Sample Rates after 60 passes through A/D and D/A Converter pairs using the system clock
There is a assertion made by some recording engineers and audiophiles that higher sampling rates are always better than lower sampling rates.
The purpose of this experiment is not to prove or disprove that assertion.
The purpose of this experiment is to examine the relative linearity of converters when operated at various sample rates.
In 2007 I set up an experiment to determine what degradation may result from multiple passes through D/A and A/D loops, and posted those results here.
As a continuation of that experiment, I decided to run some tests to determine the cumulative effect of various sample rates on a signal after multiple conversions.
For this test, I used two different converters, one based on an AKM5393 96k capable 120dB s/n chip, and one based on an AKM 5394, a 192k capable 123dB s/n chip.
The audio sources used were 24 bit 44.1kHz files.
For the tests at 96kHz and 192kHz I simply changed the session sample rate to the desired rate,
and then bounced an additional pass with no processing or converters in line at 44.1 in order to produce a file with the proper header information to play back at 44.1kHz.
This means that the audio was running at roughly 2x and 4x normal playback speed for those tests and restored to the original rate for the results.
No sample rate conversion was used.
It's important tp note that the source files were already band-limited due to the anti-aliasing filters applied during recording,
and thus they present a highly "legal" signal at the higher sample rates, and unlike signals originally recorded at those rates, cannot present any possiblility of aliasing.
Here is a link to the original source file:
Below are links to the files resulting from the experiment:
DRUMS: Converter #1, 60 generations @ 44.1kHz
DRUMS: Converter #1, 60 generations @ 96kHz
DRUMS: Converter #2, 60 generations @ 44.1kHz
DRUMS: Converter #2, 60 generations @ 96kHz
DRUMS: Converter #2, 60 generations @ 192kHz
These tests were run using the internal system clock, the results of an additional test using an external word clock are available here.
Chuck Zwicky
October 2013
All content ©2013Chuck Zwicky