Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site angband.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!ut-sally!mordor!angband!sjc
From: sjc@angband.UUCP (Steve Correll)
Newsgroups: net.audio
Subject: Re: Re: Digital Remastering
Message-ID: <61@angband.UUCP>
Date: Fri, 31-May-85 14:20:25 EDT
Article-I.D.: angband.61
Posted: Fri May 31 14:20:25 1985
Date-Received: Sun, 2-Jun-85 07:27:51 EDT
References: <356@lcuxc.UUCP> <> <337@cubsvax.UUCP>
Organization: S-1 Project, LLNL
Lines: 40

> When you copy
> (play) a digital master, isn't there a finite proability of mis-copying a
> bit?  I mean, even if parity bits are used, there's a finite probability
> of mis-copying an *even* number of bits, and no matter how many double-
> checks one cares to make, there's always a *finite*, though small, prob-
> ability of a miscopy getting through all of them, it would seem.  Thus, isn't
> it incorrect to state that "you lose nothing"?  I mean, think of how many
> bits there must be on a CD.  (OK, how many?  I'm sure many of you know!)
> Even with a very low error rate there must be a few that have been mis-
> copied... and if you know what the error rate is, and also a ballpark figure
> for what percentage of bits would have to be wrong for an audible effect,
> one could calculate how many generations of copies would be necessary to
> audibly degrade the signal -- I agree it's going to be a big number!
> (Assume random distribution of wrong bits....)

You're right that there's a finite probability of degradation whenever you
copy a digital signal.

The degradation mechanisms for digital and analog are qualitatively
different, however.  Copy an analog tape through a channel which
increases the amplitude of the signal by 5%, and after 100 copies, your
signal will be 1.05 ** 100 times too big. But if you know that the
information on the tape is a digital bitstream, you can in theory copy
the signal without error: whenever the signal is less than 50% of the
maximum, you reduce it to the "0" level, and otherwise you raise it to
the "1" level so the 5% corruption vanishes on each copy. (This digital
capability is called "noise margin".)

Of course, other degradations affect analog and digital equally.  A
tape dropout can destroy a bit completely. But compared with analog
techniques, digital techniques make it much easier to spend bandwidth
to buy error-resistance. The more redundant information you supply when
encoding a signal, the lower the probability of an uncorrectable
error.  (One reason for the enormous number of bits on a CD is the
presence of vastly more redundancy than a simple parity scheme would
provide.) But, as you have observed, the probability is still
non-zero.
-- 
                                                           --Steve Correll
sjc@s1-b.ARPA, ...!decvax!decwrl!mordor!sjc, or ...!ucbvax!dual!mordor!sjc