Well, I've had my fun, now I invite you to do the same, if you'd like. Twelve FLACs in these two zip archives…
Part One (01-06)
Part Two (07-12)
…keep in mind that each 'iteration' is actually two: once to 320 CBR M4A and then once back to 320 CBR MP3. 0 iterations is the first generation 320 CBR MP3, right off the ALAC ripped from the CD. The twelve files have been randomized. So, not in this order, the files are: 0 iterations, 1 iteration, 5 iterations, 10 iterations, 25 iterations, 50 iterations, 100 iterations, 250 iterations, 500 iterations, 1000 iterations, 2500 iterations, and finally 5000 iterations. The big guys should be pretty obvious — being reencoded 10,000 times does, in fact, take a toll on lossy audio. But the results are still pretty impressive, and I'll be curious to see if people can pick out the lower ones.
This is just for fun, of course… so I hope some of you will enjoy it. Just to reiterate, the process was to use a shell script to run XLD in CLI mode to convert an MP3 to M4A, then use afconvert to convert the resultant M4A to ALAC, then use XLD again to convert the ALAC to yet another MP3, and repeat infinitely until I finally mashed ctrl-c. All are 320kbps, max quality, CBR. The audio is an approximately 45 second sample from about 1.45 to about 2.30 of the first movement to Carl Orff's
Catulli Carmina, from the stereo Eugen Jochum: Deutsche Oper Berlin Chorus & Instrumental Ensemble recording on Deutsche Grammophon, 449 097-2.