I have been re-encoding much of my 3.90.3 APS encodings to 3.96.1 V2 (or PS). Why? Much of my music enjoys about a 10% file size reduction moving from 3.90.3 to 3.96.1 - with (as far as I can tell) the exact same audio quality.

Then I started re-encoding my Beatles collection. Wow. 3.96.1 is not nearly as efficient as 3.90.3 when it comes to The Beatles. For example, the Magical Mystery Tour CD inflates from an average bit rate of 196 with 3.90.3 all the way to 213 under 3.96.1. Pretty large increase.

Anyone know why old recordings such as The Beatles inflate so much under 3.96.1 vs. 3.90.3? I wish I had not wasted my time re-encoding these!

It sounds like you are transcoding, which could explain bloated bit rate due to noise added during the intial encoding.

No, this is a re-rip via EAC from the original CD.

I also noticed this bitrate bloat on Indian Karnatic Classical songs. For example, song 1: 203kbps vs 220kbps. Song 2: 222kbps vs 245kbps.
Encspot showes quality 78 for 3903 vs 77 for 3961.
Another observation: The scalefac is 0.2/0.3 for 3903 vs 10/11 for 3961.
This effect is similar for aps & apsy.
Any explanation ?
After experimenting with one album, I am sticking to 3903, apsy.

--Rigapada

On Beatles this is probably because of the strong stereo separation.
But this is not a huge bitrate increase...

The quality index isn't very reliable though.

based on my testing, LAME 3.96.1 saves you a lot when it comes to pretty recent CD's...especially those fully digitally produced CD's need much less space over LAME 3.90.3. Anyway, overall LAME 3.96.1 still needs less space than LAME 3.90.3

It might be that the analog noise underneath the music is inflating the bitrate. Musepack is notable for behaving the same way.

Check to see whether or not there's a correlation between bitrate bloat and analog/digital mastering.