> > From: Andrew Hobgood <[EMAIL PROTECTED]>
> > Subject: Re: MD: Found a CD that cannot be digitally recorded on MD
> >
> > | No, the LSB gets dropped and you still end up with a signal that will
not
> > | overflow the register. All that happens is you lose the extra 1 bit
> > | resolution from the LSB end.
> >
> > And this is significantly different from "clipping" in what way? Loss
of
> > data is loss of data, no matter how you try to spin it.
>
> Well, clipping manifests itself as chopping the MSB (bit or bits,
depending
> on the severity of clipping) because the incoming signal has a range far
> beyond the sampling range of the A/D converter. Magic's LSB clipping
would
> occur only if you're transferring digital data, and the frame copy routine
> is specified to maintain the MSB when discarding data. A poorly designed
> routine would accidentally discard the MSB instead of the LSB, resulting
in
> a behavior that sounds much like clipping, but without the A/D stage that
> clipping is historically characterized by.
>
> This is all IIRC, so if I'm wrong, feel free to correct me politely. =P
<Puts on polite correction voice>
Remember that 'clipping' is not the same as just losing a bit of data.
For example, say we had a three bit data system. We want to store the value
1000 (i.e. one more than the maximum, 111). If we clip the sample in the
wave that 1000 /would/ represent, could it be stored, we'd get 111 - the
nearest value we can manage. If we truncate the binary value 1000, we'd get
000 (by loosing the Most Significant Bit - MSB) or 100 (by loosing the Least
Significant Bit). Not, in sound terms, nearly as nice (but a lot simpler, I
suppose).
Anyway, all this is academic, as TosLink should transmit the 16-bit data
exactly as it receives it from a CD - it's /impossible/ (not just
improbable, but impossible) that what is being discussed on this thread is
caused by any form of 'clipping' or truncating of the data from the CD.
Jamie -> who hasn't made a post for what seems like ages (I'm still here
though!).
[The man with no sig]
P.S. the idea floating around about sign bits is probably also a red
herring - there are MANY different ways to store a sign in binary, [and none
of them is the 'right' one, though some are easier to handle than others].
Anyway, I would imagine (though I don't know - perhaps I need some polite
correction too?) that MD uses a straight 16-bit positive binary number to
store data, and doesn't dabble into any of this negative nastiness.
-----------------------------------------------------------------
To stop getting this list send a message containing just the word
"unsubscribe" to [EMAIL PROTECTED]
