Hello. cdparanoia seems to need following additions: -Rip from sector N to sector M -Skip TOC reading I'm now sure that cdparanoia is failing. Remember, the output was about the following: (== PROGRESS == [ +!---------------| 159332 00 ] == :^D * ==) The whole end is full of jitter errors, but in this case it merely must be that cdparanoia ruins the good CD audio data. The jitter correction buffers went bad at "!" point, and the jitter error cumulates to the end. So, cdparanoia algorithm should be changed to a non-cumulative. This happened when I started ripping after the "!" point: (== PROGRESS == [ | 159332 00 ] == :^D * ==) No jitter errors whatsoever. New algorithm could rip the data in several independent parts, using jitter correction on each part. Parts should overlap slightly but no jitter correction is yet applied between parts. If a part has an error, reading of that part is stopped immediately. At next round the parts which had errors are further divided to smaller parts. Then finally jitter errors are verified between the overlapping parts. This alternative algorithm uses more disk space because the parts are stored independently to the disk. Regards, Juhana