File Export menu in CS6 does not work properly?

Question

Two main issues:
1. If I downsample a 96kHz audio file to 44k1 for example I have the option to impose an anti-alias low-pass brickwall filter in the export path or not. This is a good idea. However as implemented in the Audition CS6 system we have, Adobe has the tick-box configured incorrectly. With the box ticked the filter is switched out, and with the box unticked the filter is switched in. This should be debugged and fixed. It makes me worry that other configurations might be wrong.
2. The defaults for the implementation of dither in the export of a file are inconsistent. The default should be that whenever audio data are changed by any processing, dither is applied in the exported file. This is inconsistent in Audition CS6 on our system and we have to go through all the setup for each export.

Answer 1

Thanks Steve.
I agree about the quality of Audition's sample-rate conversion. It is very good indeed and one reason we bought Audition. I also agree about the perverse logic of Audition having a tick in the box to turn the low-pass filter off, rather than on. This is is counter-intuitive, even if the excuse is that it is how it has always been historically in Audition or maybe even CoolEdit (which I have never used). Maybe Adobe could consider retitling this function so the logic is more obvious to newcomers. The riddle of how to combine requirements for 96k/24 high-res masters for one market with the needs for 44k1/16 delivery for CD, downloads and streamers remains a riddle if we are not to make our 44k1/16 masters (the main market) sound inferior to having made them 88k2 or 44k1 originally.
As for dither. My experience has been that one stage of unintended truncation does typically approximately five to ten times as much damage (irreparably) to a clean recording as does one stage of unintended dithering. I am only talking about default settings which can or should be easily overridden either in 'settings' or 'preferences' somewhere within Audition, or overridden on any single or multiple instance of exporting. The mathematics are against you on this one. If you have a 96k/24 original file and want to export a straight 96k16 file or 44k1/24 or 44k1/16 version, for me dither is obligatory. The same if you are exporting a 96k/24 file after processing. Whatever happens next. If you are exporting 96k/32 then maybe, I stress maybe, no dither is ok. Otherwise you get truncation errors, with a shower of unwanted side-effects and squeaks. That is one problem with digits rather than analogue. With digits you don't get extra random relatively benign noise with repeated undithered generations of copying, processing, transfers. The artefacts are much better described as errors and they are a shower of correlated noises. They are programme related, clock related or a combination of both so the ear is much more likely to pick them out as something less benign than two bits of pk-pk flat dither.
I understand the effects may depend on the programme material, but even two bits of pk-pk flat dither of a 16bit stream is typically 16dB to 20dB below the broadband background noise of a classical chamber music recording in a typical real-world venue. With commercial music the dynamic range tends to be compressed, so the dither will be even less perceptiable because there are fewer extended quiet sections of programme. I would prefer the effective noise level to increase by even a dB or so, if the alternative is gradually making a violin sound more and more like breaking glass from the effects of truncation.
We have been doing listening tests on dither in association with the development of new high-res delivery products. It has been interesting that no listeners detected dither as a noise 'problem' during any of the tests - even when we tried adding a lot more than 2bits' worth. In general either one noticed no change or a slight improvement subjectively with more than 2bits' worth of dither. The only negative thing we detected was that adding multiple layers of so-called random dither sometimes brought attention to the fact that it was sometimes not quite as 'random' as it should be.