[Prev] Thread [Next]  |  [Prev] Date [Next]

[SurroundSound] Re: PS3 SACD rips Hardlok Cafe Fri Sep 02 13:00:09 2011

On Sep 1, 10:48 pm, Joe A <[EMAIL PROTECTED]> wrote:
> Sorry for the long delay, my internet service was taken out by
> Hurricane Irene.
> > @Joe A -- Your statement "Therefore it is not interchangeable between
> > domains and is more akin to a lossy conversion than it is to up/
> > downsampling in the same domain of PCM."
> > May I ask, is this a definitive conclusion in some AES paper or
> > something? I find it hard to understand that converting between two
> > completely different methods of recording (and the digital formats
> > that store the information) would be akin to a lossy conversion.
> If you are familiar with video processing, there is a good analogy
> that comes to mind -- color spaces.  In video processing things are
> usually expressed in some variety of the YUV color space -- although
> native computer image processing is done in the RGB color space and in
> the print world it's the CMYK color space.
> This may not be the most appropriate analogy but I think it will help
> make my case.
> YUV and RGB images are not (losslessly) interchangeable and
> information is lost going from RGB->YUV.  They are two different
> conceptual domains, two ways of recording the same information.
> YUV saves bandwidth by eliminating color information, taking advantage
> of the fact that the human eye is more sensitive to detail than it is
> to color; it gained widespread use first in the analog world for color
> broadcasts and persists in today's digital video signal processing.
> Which brings me back to my original point -- DSD and PCM are different
> domains of audio signal processing, as RGB and YUV are in the video
> realm, and doing conversions between these domains is not
> interchangeable and not analogous to up/down-sampling within a single
> domain.
> AES is an encyrption standard, by the way...

Hi there
Glad you're safe and hope you didn't suffer too much inconvenience
from by way of Irene.

I get your argument, and as an ex-publishing person, I understand the
RGB and CMYK color space differences.
However, as pointed out by others in this thread, we're really
comparing apples to oranges when comparing high definition audio
formats, aren't we?

DSD attempts to capture an analog input faithfully, and has a wide
freq response and is considered a mastering format. PCM at 24/96 or
higher is also able to capture analog signals very faithfully. Both
apparently have some weaknesses and strengths. But both are the best
technologies we have so far. So when PS3 ripping starts to make the
DSD format accessible to more people, I guess there will be
comparisons. But to consider conversion from one high def format to
another as "lossy" sounds to me like a pointless, fruitless exercise.

There may be losses, but the losses are not intentional for the sake
of reducing file size. The losses are a natural process of converting
one information space into another type of information space. The
losses may be audible, or they may not. The term "lossy" seems to be
thrown around more often to be a synonym of "loss of information", but
before lossy compression methods like mp3 came into prominence, the
term was almost never used. In mp3 compression, the term lossy refers
to two highly specific intentional goals:

1) the irreversible reduction of digital data in a file to reduce the
file size (since lossless compression can't reduce file size beyond a
certain point of entropy)
2) the use of psychoacoustic algorithms to increase the amount of data
that can be discarded, without major audible consequences

In the case of conversion between PCM and DSD, which does not adopt
any of the above goals, losses in audio quality are just natural
losses or maybe even just attenuations of audio fidelity beyond a
certain threshold. I would go so far as to concede that conversion
between the two formats results in certain irreversible losses in the
digital audio information. But calling the process lossy would
explicitly mean that, as in the case of mp3 compression, information
is intentionally thrown away for the sake of file size reduction, at
the expense of audio fidelity (since low bitrate mp3 compression
results in weird psychoacoustically obvious effects).

Anyway, maybe it's just symantics, but I really do care about the way
the term "natural change or loss of digital data" is being replaced
with a far-reaching term like "lossy". I remember how horrified I was
when I first heard the result of severe lossy compression when I used
an mp3 compression tool to convert a 45MB flac file to a 3MB mp3 file
at 112kbps kekeke!

Sometimes I wonder... when a musical performance is recorded to
digital at 24/192, and is then resampled to 16/44.1 for sale on audio
CDs, how much audio quality is sacrificed. Can this process be called
lossy? Is it less or more "lossy" than if the original recording was
done at a lower rate of say, 24/88?
Yet, when the consumers of that CD then rip the digital audio to flac
to play on their media player, it becomes the current reference file.
Any conversion of that file to any other audio format would have to be
"lossless" if it is to be a credible point of reference to serious
music collector. As such a collector, I shy away from any music that
is in a lossy audio format, because I do not want to collect music
that has been mutilated by intentional discarding of sonic nuances,
however "below the audible threshold" they might be. Yet, up until
recently, I was already satisfied with redbook audio as long as I did
not have access to the reference-quality 24/xx studio master (and thus
did not know what I was missing). It's the same with colour or video
rendering: even if CMYK cannot boast the color gamut of RGB, most
people wouldn't called CMYK a lost cause because most people wouldn't
have seen the RGB version, and the CMYK version (if done with good
inks and good dithering or lpi on good paper) looks just dandy
anyway.  :)

You received this message because you are subscribed to the Google Groups 
"SurroundSound" group.
To post to this group, send email to [EMAIL PROTECTED]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at