Re: What's the use of a 192 kHz sample rate?

On Thu, 15 May 2008 10:16:01 -0400, "Arny Krueger" <arnyk@[EMAIL PROTECTED]
>
wrote:

>"Eric Jacobsen" <eric.jacobsen@[EMAIL PROTECTED]
> wrote in message
>news:3vlj24l9e1fb1m703e900nbqumbmibdfd6@[EMAIL PROTECTED]
>> On Mon, 12 May 2008 19:42:31 -0400, "Arny Krueger"
>> <arnyk@[EMAIL PROTECTED]
> wrote:
>>
>>
>>>
>>> A properly-operating CD player playing discs in good
>>> condition probably has less than one erroneous sample
>>> per full side (72 minutes). There are a modest number of
>>> correctable and corrected errors.
>>
>> Then they were mighty foolish to add a lot of
>> interleaving and FEC to correct the errors.    Those
>> foolish, foolish engineers.
>
>Your response is completely disconnected from what I said.
>
Not at all, but you snipped some of the relevant parts, like this note
from you:

Arny wrote:
[
"Eric Jacobsen" <eric.jacobsen@[EMAIL PROTECTED]
> wrote in message
news:66me24pdgc4m5jsd4ah01oiqli06cscvj9@[EMAIL PROTECTED]
> On Sun, 11 May 2008 16:03:56 -0400, "Arny Krueger"
> <arnyk@[EMAIL PROTECTED]
> wrote:
>
>> If the FIFO is overrun or underrun, then there would be
>> lost data and very obvious tics and pops. Obviously,
>> that has never happened in well-designed products.
>
> It probably happens more than you realize, but you may
> not recognize it.

Spare me the BS.

> Especially in automotive players, the harshness of the
> environment can make reading the disk more difficult than
> in the much tamer environment of a stationary home or
> office.

You are talking out of the back of your neck. :-(

> I've long been impressed that the CD player in
> my truck very seldom makes audible skips even when
> banging around gullies off road.   I've been in other
> vehicles that "skip" routinely in such environments, so
> there is still quite a bit of difference between various
> vendors.

> Optical disks have a fairly substantial read error rate
> even when things are going well, and introducing lots of
> vibration and harshness just makes it worse.

A properly-operating CD player playing discs in good condition
probably has 
less than one erroneous sample per full side (72 minutes). There are a
modest number of correctable and corrected errors.

>   There is
> still a good probability that the error rate or just
> plain outage duration in a player will cause the buffer
> to underflow, and there are strategies to deal with this
> in case the system manages to catch up.   The simplest is
> to just repeat the last codeword a few times, and
> sometimes that works well.   I think how well it works
> may be data dependent, but given how seldom one hears
> artifacts in players these days I think whatever
> strategies they're using are working.

Spare me the uniformed pontification.
]

So, I was responding mostly to your ad hominem response, although I
had snipped the most offending parts.

And to the point, how would you know if the FIFO hit underflow or
overflow?   Do you know what it sounds like?   For every player? 

That was pretty much my point, that different vendors have a number of
strategies to pick from, some of which work better than others.
Especially in audio the output bit error rate can be fairly
substantial and still not cause a noticeable degradation in quality. A
good experiment would be to take a digital audio stream, add perhaps a
Pe = 10e-3 bit error rate, and see whether you can tell the
difference.   Change the bit error rate until you CAN tell the
difference.   I think you'd find your assetion that a player "probably
has less than one erroneous sample per full side (72 minutes)" is
quite far from the mark.

>If you like to just make disconnected posts, please don't quote mine. 

I was responding to you.    Should I quote Shakespeare?

And it wasn't disconnected, I think you just failed to recognize the
connection.    You said:

>There are a modest number of correctable and corrected errors.

What's "modest"?   This was right after you asserted the output
"probably has less than one erroneous sample per full side (72
minutes)".   At 44kHz x 16 bit words that's a word error rate of
~5*10e-9 and a bit error rate of ~3*10e-10!  (Assuming a single bit
error.)   That's either the most over-engineered audio system around
or the error rate is much higher than you think (hint: it's the
latter).    The raw error rate on optical media, even in pristine
condition, is usually fairly high, in the Pe = 10e-3 region.    To me
that's not "modest", and is the point to which I was responding.


Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org