[Audio Empire]

On Sun, 11 Dec 2011 14:08:58 -0800, Chuck Finley wrote
(in article ):

I bought an Escient music server several years ago when hard drive space was
still relatively expensive. Most of my music on there is encoded at 320 and
192. Reviews of DACs typically discuss how they can improve the sound of CD
quality or hi-rez music, but I'm wondering what effect they would have on
compressed music. Would some kind of up-sampling device have to be added to
the DAC for this? Thanks.

OK. 192 and 320 KBPS are considered essentially "transparent", and , as far
as I can tell 320 actually is transparent to my ears. 192, OTOH, is
essentially transparent ON SPEAKERS, but on headphones, I can hear the
artifacts. They are essentially the same artifacts that I hear at 128 KBPS
and lower (noise bursts accompanying solo percussive sounds such as piano and
acoustic guitar) only much more attenuated and of shorter duration.

I use a Logitech Squeezebox Touch to listen to streaming internet radio (the
only MP3 I listen to. My CDs are ripped using Apple Lossless Compression
(ALCS) on iTunes) and the digital output of that is fed to my 24/192 DAC
through a Sonic Frontiers D2D digital up-converter (to 24/96). Basically, the
difference between the upsampled and non-upsampled MP3 audio is very subtle.
On a direct, blind comparison using the upconverter's bypass switch and a
friend doing the switching while I wear headphones in another room, I can
hear a difference when he switches but I can't honestly say that one sounds
better than the other, just "different", and frankly I can't even tell which
is which. But to the main point, the one I think you are asking, no,
up-sampling does not eliminate any compression artifacts that might be
present. If they're audible before upsampling, they're there after
upsampling. The main thing to remember about upsampling is that it adds NO
new information to a digital bit stream. It's only advantage (if any) is to
move the sampling filter cut-off from 22.05KHz (given 16-bit/44.1 KHz
sampling rate) to 48 KHz (assuming we're upconverting to 96 KHz). There is
still no information above 22.05 KHz in the reconstructed audio signal.

I hope this answers your question.