I think Pipeline is saying that you can assign the sound going out of RB (your live mix) to be recorded as the input signal on other tracks. Arm 2 new MONO tracks or a stereo track, set your recording input to be what you are hearing via your sound card mixer (or maybe even the Windows mixer on some systems). So what you hear is what you record.

It will take longer than rendering (it has to actually 'play' the song through and record it back at the same time).
But you should end up with a stereo recording containing what you hear during playback (vs Rendering) on the newly recorded tracks..

Personally I haven't ever noticed this issue before. RB renders what it plays pretty effectively, unless you are listening to 24 bit track recordings and rendering to a 16 bit wav or something .. though I doubt that is the issue.

Make your sound your own!
.. I do not work here, but the benefits are still awesome