Do different CD transports sound different?

I've recently decided to upgrade the audio system in my home office -- after all, I now spend an awful lot of time there. For years I've had a Denon all-in-one system which, to be fair, sounds pretty good. Still, there's room for improvement -- especially in the headphone output.

So I was wondering what I might use to play CDs until I get a new player/transport. Like many people of my generation, I have thousands of CDs, and I still want to play them. It occurred to me that I could use my XBox which, after all, has a CD slot and some (awful) software to play CDs. Although this does work, it doesn't sound wonderful -- the new equipment has revealed just how non-wonderful it actually is. The audio output from other sources on the XBox sounds just fine -- it's just CD playback that is unsatisfactory.

The XBox has an optical digital audio output, which I've connected to my new DAC. So (I reasoned) I'm just playing a stream of bits. A bit is a bit, isn't it? Why should this form of playback sound so awful?

This question led me -- once again -- down the audiophile rat-hole. There's no doubt that people with more money than sense are willing to hear a difference between one CD transport and another, despite a lack of credible science to back up such an effect. I'm a fairly fussy listener -- even in my office -- and I don't mind paying more for a better result. But does spending more actually get a better result? And how much better?

I should point out that I'm talking about CD transports here, not CD players. A transport just has a digital output to the DAC, while a player will have a DAC of its own. There's no question that different CD players, with their own DACs, sound different. But, bits being bits, different transports should not really sound different.

I also have to point out that I'm assuming good source material -- CDs in good physical condition. There's no doubt that some CD transports deal with defective, scratched CDs better than others. Even the humble XBox can play scratched CDs that my old Denon spits out -- that's just the result of ten year's technological development between the two. I'm really interested in whether one CD transport sounds better than another, with CDs that are in good condition.

The usual explanation why one CD transport might sound better than another is jitter. The problem, you see, is that the S/PDIF interface usually used to connect a CD transport to a DAC is inherently synchronous. It is, after all, an old technology. The rate at which data is sent by the CD to the DAC essentially sets the sampling rate of the DAC, on an instant-by-instant basis. "Jitter" is timing errors in the process of digital to analog conversion. Jitter may arise because of inherent limitations in the DAC (which the CD transport can't do anything about), or because of timing errors in the way the data is collected and transmitted.

In a CD transport, jitter might arise because the rate at which data is read from the disk is not constant, or because the data is not clocked out to the DAC accurately. Reading a CD is not a trivial task -- it requires highly accurate placement of the scanning laser, and the ability to correct errors in real time, whilst maintaining a constant data flow. Nevertheless, we do have the technology to do this, very effectively, and at reasonably low cost. Whatever jitter is present, we expect it to be small, even with relatively inexpensive equipment.

This raises the question -- how large must jitter be, to be audible? This isn't an easy question to answer. Jitter doesn't sound like noise, or even like regular harmonic distortion. The effect of jitter is most detectable (with measuring instruments) on signals that are slewing quickly. This might be because the audio waveform is of large amplitude, or of high frequency, or both. It's reasonable to assume, therefore, that its effects will be subtle, and hard to describe.

It's only in the last few years that we've started to analyse and quantify the jitter in CD transports in a reasonably scientific way. The phenomenon is certainly there, and it differs between transports.

So -- jitter is a real thing, and different CD transports experience different amounts of it. Therefore, different CD transports sound different. Case closed.

Not so fast. It turns out that modern DACs -- at least those of reasonable quality -- will reconstruct a stable clock signal from a potentially jittery input source. To some extent, they must do this, because DACs have to buffer at least enough incoming data to form a 16-bit word (because CDs use a 16-bit sampling process). So the rate at which bits are turned into voltages by the DAC is not directly related to the stability of the transport's clock and motor.

Moreover, decent DAC have greater buffering capacity than 16 bits. Many can buffer for a second or more. With this amount of buffering, the connection between the CD transport's clock, and the DAC's output, is essentially broken. So long as there isn't enough jitter to drop a whole bit -- which would be hugely audible -- the transport is effectively decoupled from the audio output.

In such a situation, if jitter plays any part in sound quality, it will be the jitter in the DAC's clock, not the CD transport.

I concede that we don't actually know how audible jitter is. It's at least plausible that it is audible to some extent, if it is not corrected. It's also plausible -- perhaps inevitable -- that jitter behaviour varies between CD transports. However, a competent DAC ought to be able to eliminate the effects of jitter on the incoming data completely. There's really no reason to expect different CD transports to sound different with a reasonable, modern DAC. And, if your DAC is incompetent, then jitter in your CD transport is probably a lesser problem.

So -- stupidly-expensive CD transports are, in fact, snake oil, as so much is in the hi-fi world. That does leave unanswered the question why my XBox makes such a poor CD transport.

The explanation, I believe, is that it isn't acting as a transport at all. That is, the XBox is not simply reading bits from the CD and sending them to its S/PDIF output. I know this is the case, because the XBox mixes the CD's audio with other sound sources. I suspect that the XBox is actually resampling the CD audio, to make it possible to do this mixing. It's certainly possible to resample audio in a way that is essentially inaudible, but it's also possible to implement resampling badly.

As ever, technical knowledge is the best defence against the snake-oil salesman.