Audio, hi-fi and car audio - Dirty Digital [sic.]

Audio Banter (https://www.audiobanter.co.uk/forum.php)

- uk.rec.audio (General Audio and Hi-Fi) (https://www.audiobanter.co.uk/uk-rec-audio-general-audio/)

- - Dirty Digital [sic.] (https://www.audiobanter.co.uk/uk-rec-audio-general-audio/7456-dirty-digital-sic.html)

Dirty Digital [sic.]

Has anyone read the article "Dirty Digital" by Noel Keywood in July's
"Hi-Fi World"? I picked up a copy for some amusement during a journey
yesterday and was stunned by the article's technical incompetence.

Keywood says of CD that "dynamic range is limited to 85 dB or so by
dither noise" (incorrect - it's actually about 93.3 dB IIRC for 2 LSB p-p
of TPDF dither). Then he complains, disingenuously, about the "dirty
distortion" of quantization on CD (he calls it "digital distortion"),
misunderstanding the fact that it just isn't present when you use dither.

He even misunderstands the issue of correlation between signal and
quantization error, calling them uncorrelated whereas they actually
are correlated and dither's job is to de-correlate the two.

He mixes up distortion and noise inconsistently under the single term
"distortion" sometime meaning noise, sometimes distortion and sometimes
noise+distortion.

Here's another example. He says CD distortion [sic.] is 10% at -80 dB
(dBFS he means, I assume) and later says that "LP is benign and also
produces ten times less distortion than CD at low levels (0.1% to 1%)".
In fact at the comparable low level on LP of -80 dB (WRT 0 dB = 5 cm/s,
I will assume) the LP noise floor is actually above the signal and the
LP's distortion in Keywood's terms is above 100%.

He says of the music business that it "now tries to keep music levels
well above the unacceptably high distortion floor of CD that we illustrate
here". Well, that's a new explanation for the "loudness war".

I think Keywood should have a good look at Jim Lesurf's "Good Resolutions"
article at http://www.audiomisc.co.uk/HFN/goodr...ons/page1.html
(in particular figure 5 on page 2 will show him a realistic comparison
of dynamic range of CD versus LP). Also Jim's article "In a Dither"
at http://www.audiomisc.co.uk/inadither/Page1.html would show him where
his misunderstanding about dither lies.

Keywood starts out with "This isn't a hatchet job of the poor little
silver Frisbee" and then proceeds to try just that. In fact through
his clear technical misunderstandings the only thing that suffers a
"hatchet job" is Keywood's own reputation for competence.

--
John Phillips

Dirty Digital [sic.]

In article , John Phillips
wrote:

[snip]

I think Keywood should have a good look at Jim Lesurf's "Good
Resolutions" article at
http://www.audiomisc.co.uk/HFN/goodr...ons/page1.html (in particular
figure 5 on page 2 will show him a realistic comparison of dynamic range
of CD versus LP). Also Jim's article "In a Dither" at
http://www.audiomisc.co.uk/inadither/Page1.html would show him where his
misunderstanding about dither lies.

I also suspect that with many of the old analogue original recordings the
recorded noise level is enough to dither adequately any transfer onto CD-A
- even if the engineers don't explicitly add in dither or noise shaping whe
doing the ADC. :-)

FWIW **many** years ago I contacted him, giving much the same explanation
as on the pages you refer to above - including example spectra - and giving
details of the basic explanations. Prompted by an article he'd published
back then making similar comments to the recent one.

I hoped/assumed at the time that he'd realise he might simply be making a
mistake, and misunderstanding the situation. Point being that what I was
saying wasn't 'new' or my own idea. Just standard information theory,
supported by practice, and evidenced by appropriate methods.

Got no-where.

I'm afraid he regularly makes comments like those in the recent issue of
HFW. I read them and just sighed.

I also tend to take with a pinch of salt some of the THD values they report
for things like CD players. Partly due to the possibility they are using
undithered discs - so reporting the effect of undithered quantisation.
Partly because the values often look to me just like the noise floor you'd
get from dithered FFTs with durations in the range 16k to 64k samples.

If so, the consquence could be that the values they publish may tell you
more about their measurement system and how they 'interpret' the results
than about the CD player being tested.

Keywood starts out with "This isn't a hatchet job of the poor little
silver Frisbee" and then proceeds to try just that. In fact through his
clear technical misunderstandings the only thing that suffers a "hatchet
job" is Keywood's own reputation for competence.

As I say above, NK has form as long as yer arm on this. :-) From comments
I have heard in private, others have also given up trying to explain some
things to him. To me, this is a shame. He did do some superb work in the
past on topics like the resonances in tone arms, etc, which showed up some
very interesting things. Deserves respect for much of what he did. I also
admire HFW for when they do things like publish kits for amps or speakers
and encourage readers to get involved in building. I was pleased when HFW
started up for these sorts of reasons, and good luck to them. But... I
guess that we all get in a muddle about some things. Only human.

So I doubt he'd read the webpages you reference above [1] or, likely - if
he did - then he simply wouldn't accept what they say.

Slainte,

Jim

[1] Perhaps worth admitting here that the webpages were themselves prompted
by seeing NK make these sorts of assertions in another HFW item a few years
ago. But when producing the webpages I concentrated on the topic and
glossed over where I'd read what I was reacting to. :-)

--
Change 'noise' to 'jcgl' if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Dirty Digital [sic.]

"John Phillips" wrote
in message

Has anyone read the article "Dirty Digital" by Noel
Keywood in July's "Hi-Fi World"? I picked up a copy for
some amusement during a journey yesterday and was stunned
by the article's technical incompetence.

This page seems like a potential sounding board for your comments:

http://www.hi-fiworld.co.uk/hfw/email1.html

BTW, your comments seem well-reasoned, and believable, although I haven't
read the article.

Dirty Digital [sic.]

Jim Lesurf wrote:

Just checked, and my copy of the letter is dated 1993. So 25 years ago!
:-)

15 years or 1983?

Chris
--
Chris J Dixon Nottingham UK

Have dancing shoes, will ceilidh.

Dirty Digital [sic.]

In article ,
Chris J Dixon wrote:
Jim Lesurf wrote:

Just checked, and my copy of the letter is dated 1993. So 25 years ago!
:-)

15 years or 1983?

1993 is correct. The '25' was a 'senior moment'. :-)

Slainte,

Jim

--
Change 'noise' to 'jcgl' if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Dirty Digital [sic.]

On 2008-06-16, Jim Lesurf wrote:
In article , John Phillips
wrote:

I also suspect that with many of the old analogue original recordings the
recorded noise level is enough to dither adequately any transfer onto CD-A
- even if the engineers don't explicitly add in dither or noise shaping whe
doing the ADC. :-)

I have wondered about that and the CD-A timeline.

IIRC Lipschitz and Vanderkooy were publishing about dither in JAES in
about 1984 and just after. Although dither had been know for a long
time I suspect you are right that noise floors for material transferred
to CD were probably sufficient in the early days of CD (1982-ish)
to render external dither unnecessary.

One of my musings was that for a short period after ADCs got better it
may have been that CDs lacking dither but displaying the effects of
quantization noise might have made it onto the market. I have no idea
if this did actually happen, though.

As I say above, NK has form as long as yer arm on this. :-) From comments
I have heard in private, others have also given up trying to explain some
things to him. ...

Well I guess I won't try.

--
John Phillips

Dirty Digital [sic.]

"John Phillips" wrote
in message
On 2008-06-16, Jim Lesurf wrote:
In article ,
John Phillips wrote:

I also suspect that with many of the old analogue
original recordings the recorded noise level is enough
to dither adequately any transfer onto CD-A - even if
the engineers don't explicitly add in dither or noise
shaping whe doing the ADC. :-)

I have wondered about that and the CD-A timeline.

IIRC Lipschitz and Vanderkooy were publishing about
dither in JAES in about 1984 and just after. Although
dither had been know for a long time I suspect you are
right that noise floors for material transferred to CD
were probably sufficient in the early days of CD
(1982-ish) to render external dither unnecessary.

AFAIK Vanderkooy and Lip****z were knowingly publishing old news, in an
effort to overcome some pretty strange false claims that were being
circulated at the time by people who should have known better.

One of my musings was that for a short period after ADCs
got better it may have been that CDs lacking dither but
displaying the effects of quantization noise might have
made it onto the market. I have no idea if this did
actually happen, though.

The earliest CD players had converters good enough to demonstrate dynamic
range on the order of 93 dB, which is pretty close to the theoretical max:

http://www.pcavtech.com/play-rec/Son...ndex.htm#DR_DA

About 15 or more years later, a highly-regarded CD player improved on the
legacy players performance by only about 1 dB

http://www.pcavtech.com/play-rec/cd67se/index.htm#DR_LB

The noise floor of a well-made recording is on the order of 75-80 dB. Below
that is the noise floor, usually from analog (thermal) sources. This is many
times more than is required to properly dither a proper 16 bit conversion.