Is it even possible to be *audibly* better? (if "better" = objectively better, as in less distortion, flatter FR, etc.)
In case of traditional audio metric (that you mentioned) relationship between perceived audio quality and objective measurements is poor and your question is understandable. In SE audio metric such relationship is more defined, so in some cases we can safely predict subjective quality from objective measurements.
Most real-life audio material (especially if it is perfectly mixed and mastered) has natural high frequency roll-off
Most real-life audio material (especially if it is perfectly mixed and mastered) has quite a lot of energy above 20 kHz (that is, between 20 kHz and the upper limit which is usually 22050 Hz), even with roll-off it's still quite a lot; and sometimes people use extreme settings for noise shaping (for 16 bit conversion) which adds to that band too. Humans, though, don't hear that band at all, so taking it into account at all would introduce unnecessary mistakes.
The test set “Variety” (2 hours of music), which is used for df-measurements has the following overall amplitude-frequency characteristic:
Not so much energy between 20kHz and 22.5kHz indeed.
Last post by jacobacci -
Thanks guys Detect indexes = true is turned on. However the disc is not listed as having pre-emphasis. Where on the screen would I see if it was? Luckily the highs were so overly bright, that it was obvious.
Last post by ManInTheDark -
Recently, I got a CD that has HDCD on it. I decoded a track from it using the HDCD decoder (foo_hdcd) in foobar2000. And...nothing happened. In fact, it had the same Replaygain level as the 16bit source file. But with hdcd 0.2? The total opposite happened! Here's the stats:
01. -8.28 (16bit) Whenever It Rains.wav 02. -8.28 (24bit) Whenever It Rains.24bit.01.wav (foo_hdcd) 03. -2.26 (24bit) Whenever It Rains.24bit.02.wav (hdcd.exe 0.2)
WTF? Obviously, foo_hdcd isn't working, and hdcd.exe 0.2 is I guess I'm better off creating a folder with BAT file, putting all my HDCD files into it and converting everything that way.
Last post by Porcus -
I can imagine that 32 bits floating-point is good for those who notoriously mess up their conversions ... and I can't help myself thinking that such things happen deliberately for the purpose of creating a "problem with digital".
Although there is even well-reputed software which cannot convert reliably from floating-point to integer.