Martijn van Buul writes: >Let me put it this way. Suppose there's an error somewhere on a CD, say >for 100 samples. The CD just can't read it, and has to interpolate, and >inserts data that won't sound too awful. A human mind won't notice >the difference, since its a) a very short error (2 ms) and b) masked. That's not how the samples are stored. Audio CDs use an interleaved Reed- Solomon error correcting code. Bits belonging to one sample are scattered over a wide area of the disk - nearby bits belong to different samples. If you get a long run of errors (which is the most likely case for scratches), you're going to lose a bit or two from each of many samples, and the error correction can deal with that very easily (Reed-Solomon is a class of codes which use redundancy to correct errors. The more redundant bits you add, the more error bits you can correct. I don't know off-hand the number of bits that the flavour used in CDs can correct). Interpolation is only used when things go very very wrong. A lot of the loss when going from audio to data is from giving the disk more structure. It's like a floppy disk - you can store a lot more on a disk if you don't bother with sector headers and gaps. There may be more integrity checks as well, but an extra 15% in raw bits doesn't get you much in the way of error correction. John West - This message was sent through the cbm-hackers mailing list. To unsubscribe: echo unsubscribe | mail firstname.lastname@example.org.
Archive generated by hypermail 2.1.1.