The methodology, along with various post test Q&As are available.
With regards to the audiophile test subjects, what was interesting is that they were allowed to do the ABX tests in their own homes, on their own stereos and in their own time - thus negating the [misplaced] criticisms of ABX testing you hear from subjectivists. None were able to pick the high res source from 16/44 at a rate greater than a fair flipping of a coin.
Bear in mind this was just one of the many tests on this subject, like as in the links below:
http://archimago.blogspot.com.au/2014/06/24-bit-vs-16-bit-audio-test-part-i.html
http://archimago.blogspot.com.au/2014/06/24-bit-vs-16-bit-audio-test-part-ii.html
Thank you for links.
1) I respect Archimago. He do many good and very interesting researches.
This experiment carefully done
in its conditions, that carefully described. Conclusions are done properly.
2)
I don't claim that 24 bit better 16 bit.
3) I think, that
comparison of 16 and 24 bit is technically impossible.
4) I consider
proper double blind test as correct estimation of
subjective audio quality (that is last criteria, want you it or don't) via
objective (repeatable result in certain conditions) method.
However, need consider some things.
By link compared playback of 2 formats of files: 24 and 16 bit.
1. There was checked many devices. But DAC use different circuits or different parts of circuits.
So this way compared implementations, not formats as itself.
2. ABX test is not home experiment.
There need big number of measurements (several thousands though for proper calculations) and
group of certified observers, that carefully control experiment performing.
Results should be fixed in the protocol, that each of observers sign.
Also need exact schemes for each experiment.
3. I don't know what to do foobar ABX plugin, when prepare files for comparison (you can see progress before it run).
4. There was not checked
depending of quality for 16 and 24 bit on
level for each of tested files.
Example:
If you test dithering and signal have high loudness, very probably dither give nothing.
If you reduce loudness into file 40 ... 80 dB and increase amplifier loudness, there, probably, difference appear (depend on dither implementation and experiment conditions).
5. Need control SPL (sound pressure level) in the listener ear point.
6. Different spectral distribution in test files can cause diffrent resuts. This issue need to learn.
7. All measurement tools should have proper precision, be certified and checked by certified organization.
Resume:
There are many subtle details for exact experiment.
16 or 24 or 32 bit is not our aim.
It is only our tools for implementations of
subjective improvemnents, that should be
proved via objective methods.
If 24 bit don't improve something in certain case, don't use it in the case.
Experiment conditions are first thing that need to know