In reply to:

---

Then again, there's a whole area of audio equipment (like the Sony ES compact disc players) which are more expensive and go to extra trouble to make sure the line-level output is perfect. But some of the extra things they do sound like snake oil to me, like using an extra-stable CD transport mechanism to reduce vibrations transmitted to the disc. I mean, you can prevent skips this way, but other than that, the data is digital and the transport mechanism has nothing to do with the sound as long as the bits are there.

---

Let me start by putting this question to you. A CD-ROM can take 650MB of data, but a CD has 72 minutes of audio. 72 minutes of uncompressed stereo 16bit 44.1KHz music is 762048000 bytes, or 726.7 megabytes. So how do they fit that on there if a CD-ROM only contains 650MB?

Simple. CD Audio is stored in an error-[u]compensating[/u] format where as CD-ROM data is stored in an error-[u]correcting[/u] format. I'm not clear on the exact details, but basically the CD Audio is stored so that bit errors eventually even out - the audio will eventually get back to the correct position even if the immediate sample might be out by a least-significant-bit or two. CD-ROM data, on the other hand, is stored in an expanded format so that bit errors can be rebuilt transparently.

The point here is that CDs usually have a couple of hundred bit errors per second of play time (don't quote me on that, those statistics are coming from old memory). These bit errors are coming from fluctuations in the disc's speed and distance from the lens, scuff marks and prints, and other interference with the laser beam during playback. The playback algorithm is able to minimise the impact of those errors but sometimes it doesn't compensate completely; or, for your audiophiles, as completely as they'd like. Consider that the least-significant bit in a 16-bit sample represents about -96dB - in other words, in a 96dB signal they represent the bottom 1dB. So while ordinarily you may not hear bit, or even two- or three-bit, errors, they still exist in the audio and therefore can affect very sensitive listening devices.

Whether people's ears are 'very sensitive listening devices' is another question entirely.

The uptical pracshot of this is that some audiophiles might want to pay extra for a device which claims to have higher quality components to reduce the chance of these defects making it through to the audio stream.

Also keep in mind that these defects will affect a CD rip - the audio is [u]not[/u] coming off the disc in the same way as a file does.

Does that all make sense now?

Save the whales. Feed the hungry. Free the mallocs.