When I initially ripped the CDs, I used the software’s “standard” MP3 setting (LAME encoder with variable bit rate (190KBs estimated) which upon closer inspection under the “advanced” button shows frequency=source, channels=auto, and quality=default).

Afterwards, when I perform a batch conversion (forcing 16 bits and 44.1KHz sampling) of the resultant MP3 files ripped with the original settings, I see an average 16% reduction in file size and no audible difference (to me) in recording quality. The waveforms in Audacity look identical. I have not yet figured out how to do a “diff” comparison of the alternate rips, but my guess is that the data inflation from oversampling is at least partially offset by compression of consecutively unchanged data elements.

So, given that the source material is CD quality (16 bit 44.1Khz) is it worthwhile to re-rip at the source rate or convert the resulting MP3s to the smaller file size resulting from 16 bit 44.1Khz?

Thanks for any advice.