on 3/19/03 6:06 PM, Mike Richter at [log in to unmask] wrote:

> At 05:17 PM 3/19/2003 -0500, Steven Smolian wrote:
>> Maybe so, but I hear differences sufficient to have made lazy old me take
>> time to investigate this.
> Fine. The investigation is straightforward.
> 1. Take a set of 'redbook' WAVs from your hard drive
> 2. Write them to a disc known to produce known errors at your write speed
> 3. Rip the tracks to WAV using your choice of program, correcting for
> offset as required
> 4. Compare the original and the extracted WAVs with WinDiff or equivalent
> NOTE: If you see occasional errors, you have a poor write in step 2 or poor
> extraction in step 3. Any change of overall parameters, such as frequency
> response or dynamic range will be continuous.

It is fairly well accepted among many mastering engineers that audio CDs can
be bit identical and still sound different under certain circumstances.

When comparing two _audio_ CDs, the preferred method is to compare the
digital audio.

Here's what mastering engineers do: use the digital output of the audio CD
player and load the two CDs into a digital audio workstation. When both CDs
are in the workstation, you time align the CDs in matching panels of an edit
decision list, and flip the polarity on one CD.

If the two audio CDs are identical, playing the EDL should produce total
silence because they cancel out each other. If there is any resulting audio
when playing the EDL, then the CDs are not bit equivalent. This procedure
eliminates needless conversion to data files with its potential vagaries.

One to One, the mastering and replication magazine, examined this issue some
time ago and reported that _in_some_CD_players_ bit identical audio CDs may
sound different because of the increased overhead caused by excessive
correctable errors on the poorer quality disc. Specifically, an audio CD
player encountering an audio CD with a higher rate of correctable errors
will attempt to reread the CD more often, causing a greater load on the
power supply at the expense of the analog circuitry. The difference, or
deterioration, is characterized by a collapse of the sound stage (loss of
the delicate spatial cues), reduced bass response, and poorer transient
response - even though the digital audio from the CDs may be bit accurate.

These differences are not as audible when a high quality CD player is used,
and may be entirely inaudible when a high quality outboard D/A converter is
used. In other words, it can occur when the analog circuitry shares the
power supply with the motor and servo for the CD player.

I once had a discussion with a matrix engineer at one of the most highly
regarded CD plants who insisted that he could hear a difference between
identical manufactured CDs, one with little or no silkscreen paint, and one
heavily painted. He said in every case he had examined, the CD with minimum
silkscreening sounded better, and he presumed it might be because 1) the
silkscreening may cause the audio CD to balance less well as it spins,
and/or 2) the additional opaque layer so near the reflective layer may
interfere slightly with the reading of the pits, causing more rereads or
error correction attempts.

Later. Parker

Parker Dinkins
MasterDigital Corporation
CD Mastering + Audio Restoration