[Audio Empire]

On Tue, 30 Aug 2011 05:18:27 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message
...

A rule of thumb that I have gleaned from
lots of experience and study is that 24-bit/48KHz quantization is
indistinguishable, audibly, from bit-depths higher than 24-bit or from
sampling rates higher than 48 KHz. Nobody's hearing is good enough to
distinguish any theoretical or practical advantage to frequency responses
that go much above 22KHz, or dynamic ranges that exceed 120 dB.

Few if anybody's recording space or listening room is good enough, either.

JJ had a listening room at the AT&T labs that was designed to support a -100
dB noise floor for practical and typical listening levels per EBU
recommendation BS 1116.. He can tell you about the slings and arrows and
costs of actually doing such a thing. If memory serves, a freeway a fraction
of a mile away was one of the hurdles that they had to overcome, all at
great cost to the management.

If 24-bit has any REAL WORLD advantage it is that it allows for lower peak
levels on recording which lessens the danger of over-modulation

Audible distortion due to approaching the point of over-modulation does not
exist in the digital domain, and only exists for the upper 1 to 3 dB in the
analog domain except for things like magnetic tape. Adding some 50 dB of
dynamic range with 24 bits versus 16 bits looks great on paper, but it does
not help with problems that small.

I think you misunderstand me. In recording, you have two opposing goals: (1)
to record peaks at as high a level possible without over-modulating
(allowable in analog recording, with occasional, momentary, excursions to
+3dB being of no consequence but anathema in digital recordings where trying
to use bits that don't exist results in nasty noise.) and (2) while
simultaneously trying to keep the low-level info in the recording out of the
mud and to do so without gain riding or using analog audio compression to
restrict the dynamic range.

without the resultant recording ending up down in the mud where
distortion and
quantization noise increase as the signal toggles fewer and fewer bits.

Quantization noise is a straw man with modern digital equipment because its
hard to do Sigma-Delta converters without introducing dither or something
like it.

Quantization noise might not be a problem with dithering, but rising
distortion certainly is a problem. Low level signals are much better served
by 24-bit than by 16. It might not matter with pop music, but it certainly
does with classical. If you don't believe me, try recording a clavichord as
I recently did.