I have long pooh-poohed the idea that CD players could sound different, but I had a chance this weekend to try a blind test; and slap my thigh, it's true, they do.

So I had two players, and everything else in the chain was identical. I was plugging a single TosLink cable into one player and then the other, using the same physical CDs, and external DACs and amps. The lucky lucky blind tester was my wife, and of course I was also listening, but not blind.

There was a distinct and noticeable difference between the players. I'm not saying one was better than the other, or getting all stereophile about <meaningless acoustic attribute of your choice>, but my wife got it right (that's player 1/that's player 2) 100% of the time.

My point (eventually, thank you for sticking with it), is why ? A CD player reads 1's and 0's off a disk and pipes them down a fibre optic line at a fixed rate. It's the DACs, amps, and speakers that introduce the tone of the sound, surely ? If two players sound different, where is that difference coming from - presumably if they interpret the disk differently one of them would be generating errors in the DAC ?

Regards

Mark

CDs are expected to be flawed, regardless of scratches. Part of the reading process is in correcting those errors. It's possible that the two drives correct the errors in different ways.

Data CD filesystems also expect there to be errors, and are encoded with a real error correction algorithm (IIRC, it's actually two or three passes of Reed-Solomon) so that you can get the exact data back off of it. Audio CD data has no such facility and expects the CD player to fill in the blanks based on no additional information -- just extrapolation from surrounding data, I guess.

Audio CD data has no such facility

To be absolutely scrupulously accurate, audio CDs have fewer such facilities -- they do use CSIRC, a cunning sort of splayed double Reed-Solomon. But wfaulk's point is still valid: if the CD is producing C2 errors (i.e. errors uncorrectable by CSIRC) then two players may well have different algorithms for filling in the gaps.

However a disc usually has to be visibly damaged to produce C2 errors. I'm thus pretty much as mystified as you about differences arising from (as your account seems to imply) any point on good CDs. The literature will have you believe it's all in the "jitter" -- the regularity or otherwise of the data clock on the fibre interface -- but it seems pretty likely that the DAC would reclock the data, so in your setup clock issues shouldn't arise, and anyway I'm not convinced that jitter is necessarily perceptible.

Peter

...and progressing that a little further;

Consider the reverence that EAC and cdparanoia hold in the digitial music world. They have gotten there on merit - that is they are better able to correct issues from scratches than just using the standard reed-solomon scheme. AFAIK, they do this by re-reading any suspect sectors as many times as they need to in order to gain confidence in the data.

Some high end CD transports also do something similar. Another factor is the drive mechanics. High end mechanics are going to track better and read the correct data first time more often than low end mechanics.

But assuming that the data is read perfectly from the disc, as far as I'm concerned, bits are bits.

Some audiophiles maintain that bit jitter in the digital stream causes differences. Given that nearly (if not) all CD players buffer the data and clock it out independantly of the rate that it as physically read, I can't see why this situation should exist. (Let's assume for the minute that we are talking about players with a 'reasonable' standard of engineering, ie, not the $25 NoName special designed by monkeys in a sweatshop somewhere).
Anyway, with a digital link, the jitter is more likely to be affected by the electronics in the receiver.

It always amuses me to see people claiming an obvious and audible benefit in using $200 coax cables for spdif connections.

but I had a chance this weekend to try a blind test

Can you describe the protocol used for this allegedly blind test, and what your actual results were (number of hits and misses)? Subtle things can cue you as to which player was which, such as different sounds they might make when loading the disc.