[sdiy] OT: Scratchy CDs

[ChristianH]

I never completely understood this, but AFAIK there is no decent
_random_access_ read command for audio data with a CD-ROM drive. Once a
program has to stop reading, in order to write the data to disk, it only
can 'go back a little bit' and start reading sequentially again. It has
to put the newly read data into relation to the previous chunk, usually by
looking for an overlapping identical data area. This resynchronization
process can be implemented well, or it can be done sloppy. If it's not so
good, it can fall for repetitive audio contents. Happens easily with
sparse techno stuff (especially if it's sample based ;-), or with quite
the opposite, long pad chords without any rhythm.

I heard rumors that SCSI drives are better in this respect, but don't
know what may be the reason for that. Might be one of those myths as
well. At least I never had a single stutter problem when reading audio
CDs using the trusty old SCSI TEAC CD-R55 burner and an age old version
of the WinOnCD software.


But what's puzzling me, is just the nature of my game - oops, I mean,
that the error correction failure affects only very short sample runs.
With floppies, it was quite common for a whole sector to fail. And IIRC,
the sample bits on a CD aren't simply coded as pits or no pits, but
instead certain pit runs are combined and interpreted. I'd expect some
bit level resynchronization problems, if there are bits missing. But
maybe exactly that was the intention for combining several bits into
runs, and by having redundancy (by allowing only a certain number of bit
combinations) this might be helpful for quick resync.

pleased to meet you, hope you guess my name...
Christian



On Tue, 23 Sep 2003 17:39:04 +0200 Ingo Debus wrote:

> 
> Am Montag, 22.09.03 um 22:25 Uhr schrieb ASSI:
> 
> > The error correction process for an audio CD is rather complicated and
> > as you found out, when all else fails, the drive will output the last
> > known good sample in the absence of any better information.
> 
> Does this error correction also work when a track is "ripped" (or 
> however it's called) via a CD-ROM drive in a computer? I was always 
> under the impression that it does not, since I got so many bad data 
> when ripping with my old computer. I always had to proof-listen before 
> burning a CD. How good the ripping worked seemed to be dependent of how 
> much the CD was scratched, but even badly scratched CDs play well on an 
> audio CD player.
> 
> Ingo