The dynamics you can perceive don't change until you can "hear" low level details being drowned in quantization noise. IIRC nobody complained about the noise floor.Anyway, my point was to show that 16 bits all the way can be "overkill" even with dynamic classical music. Bit depth can be reduced much further with modern compressed pop or metal to like 8 or 7 bits - still being almost indistinguishable from the full 16 bits.24 bits ... yep, totally worth it, NOT.
I like these kind of experiments now and then but prefer to just listen to music and enjoy it.
I can easily manage that with a Sansa Fuze + KSC75 b.t.w.
Having said that, I do 'believe' 16 bits is enough for playback (not for recording, manipulating, processing, filtering or when digital volume control is used) and to prove 24 bit files is overkill compared to 16 bits a better test might have been conceived.Perhaps by reducing an original 24/96 to 16/96 and convert pack (with bit padding, no interpolation) to 24/96 and compare those files.Frequency range and dynamic range would have remained unaffected and 16 vs 24 bit would be the only 'change'.Only by nulling and viewing the resulting noise levels/spectrum clues might be found if one was to cheat by analysing.My guess is that even the golden-est of ears would not have been able to determine the difference with a high degree of certainty on even the most revealing equipment... I could be wrong though...
What are you talking about? Most of the listeners mentioned that the A sample sounded cleaner.Also, where did you prove your point? Almost all participants managed to notice differences between the two samples with most liking the unmolested sample better.It seems that you have not even bothered to carefully read & correlate the impressions of the participants & that you had already made up your mind as to what the result was going to be and didn't bother to process the actual data that came out.
Sure I did, not a single successful ABX log was posted. Gut feelings really are great, everyone has one. Usually heavily biased.
I'm not saying nobody could pick out any kind of difference. I deliberately chose the bit depth "high" enough not to cause blatantly audible differences but low enough to not make distinguishing them impossible. Raising the bit depth by maybe 1 to 2 bits (into approx. perfect vinyl territory) and you'd have had even less success in your ABX and the others less "luck" with their gut feelings.When I would have to re-do the test I'd only change one thing: less aggressive noise shaping (which resulted in a slight boost in the FR in the highest octave, again biasing those who'd look at a spectrum analyzer first).
