CD Read Errors--Rare?

Music is encoded on a CD in digital form as a stream of binary digits or bits. A typical CD may have as many as a million errors.

Here's the best and relatively non-technical explanation of why that's not a problem:

https://www.irishtimes.com/news/science/high-fidelity-how-the-sound-of-cds-stays-error-free-1.2362987

Consider an array of 4 disk drives, a Redundant Array of Independent Drives or RAID.  I write data on the first 3, and on the 4th and calculate the parity (parity is calculated from the sum of the data bits, and is either even, 0, or odd, 1. So, writing 001 sums odd, so the parity is 1, writing 011 sums even, so its parity is 0). When I read the data, if the read sum doesn't match the parity, say 011 is misread as 010, then the parity check algorithm determines the error and corrects it. 

N.B. Modern disk arrays, as used by a streaming service for example, contain hundreds or even thousands of drives and might use striping across as many as twenty or more drives with double parity bit error correction built in. They continue to operate flawlessly even when one or two disks fail completely.  And with a thousand or two thousand drives per array, individual drive failures are common. 

 

 

It wasn't the single bit error he was concerned about if I remember correctly. It was the two bit error, which you can definetely hear, even on LPs.

Hartley's claim was that even though the sample is only 1/44100 of a second long, if the error occurred at a point where there was a major shift in the sound (my words) it would be translated as a click. The click would be audible. If that is the case, the question would be ... how often do your hear clicks on CD playback? I can't remember hearing even one anytime in recent memory.

Regards, Rich

The error correcting code used for CD data is not a band-aid for the occasional read error. It is an essential part of the data storage and retrieval process. It permits the system to be operated at a much higher bandwidth than the hardware could manage if read errors could not be allowed to occur. Read errors are frequent, but a group of errors extensive enough to get through the error correcting code is rare.

I don't remember if it was the latest TAS, but I recall Robert Hartley talking about potential CD read errors, and stating that they should be rare. I think the reasoning was much deeper than that we would hear clicks if there were read errors. There was a discussion of the error correction encoding and he detailed how many consecutive bits would have to be misread and/or unable to be read, to generate a genuine read error.

Reading off a CD for playback should be basically error free. Reading off a hard drive or out of RAM would be literally effor free. It just seems that the handling of digital files in the digital domain is a solved problem.

It depends. There are a lot of effects going on. Nerve bundles in the ear fire slowly - less than 4 Khz. So one individual error may or may not be audible depending on when it occurred. Of course many errors would audible.