In article , John R Leddy
wrote:
'Jim Lesurf[_2_ Wrote:
;94178']The problem is that real DACs aren't perfect. So when
replaying 44.1k a DAC may generate effects which extend well below
22.05kHz.Jim,
Even though a DAC or player may use an LED or display to indicate the
file format entering the component, isn't it possible the DSP could be
upsampling anyway?
Yes, quite likely in fact. But if the original was downsampled to 44.1k
that downsampling would introduce 'flaws' to some extent, and so would
upsampling again in the DAC. These 'flaws' may be inaudible (particular if
you don't have the original to compare with). But may not.
From the POV of preserving quality and minimising the addition of
flaws/losses that may become audible the ideal is to avoid all conversions
that can be avoided.
Given that it is now easy to record and playback 96k/24 this means it is
now easy to simply avoid and forget such 'risks' as downsampling and
upsampling along the way.
'Jim Lesurf[_2_ Wrote:
;94178']Simple plots of what single-frequency tones someone can hear
don't tell you about what happens when they hear more complicated time
varying signals with many components. Look on the web for the work of
Oohashi for example. Human hearing is non-linear. Real DACs, etc, are
non-linear too. The presence/absence of 'Ultrasonic' components can
affect what we perceive. Depends on the circumstances, etc.
Would I be right in thinking we are all continuously bombarded with
frequencies above and below our hearing capabilities
Yes. However the point to bear in mind is that for, say, acoustic
instruments, the 'ultrasonic' output is coherently related to the 'audible'
output in a way characteristic of that instrument. This isn't the case for
random background noises like someone dropping a bunch of keys some way
away from you. Human hearing, being *physically* non-linear can therefore
produce different results for the cases.
How much this *matters* is another question. But there is reasonably well
obtained (in academic terms) evidence that it has an effect on human
perception.
, and as such this
condition would simply envelope ultrasonic frequencies deliberately
introduced to an audio replay system, to the degree we wouldn't
experience a difference if those ultrasonic frequencies were absent
either?
No, the results may be different for the above reason.
In general terms a linear swept tone burst has the same spectrum as an
impulse function with the same energy-time spectrum. They sound very
different, though. Timing and coherence do matter.
What with memory, my brain manipulating my own high and low pass
filters as I concentrate, and the varying noise floor within my
listening environment, is it realistic for me to be concerning myself
with something imperceptible? I may have just answered my own question.
As you realise, by self reference if you find it is 'imperceptable' then it
won't matter. But it is quie possible that in some cases it *is*
perceptable.
However, whilst I'm not inclined to investigate this further via the
web, I would respect your opinion.
'Jim Lesurf[_2_ Wrote:
;94178']So the advantage of playing 96k material on a good system is
that it tends to shove these problems further away from the regions
where they can have an effect on what we hear.
Handy in the studio for minimising additional noise when utilising
multiple effects, but once packaged for replay haven't the benefits
already been achieved?
Not if you then downsample for final delivery to the playing system. That
then can introduce flaws/changes
'Jim Lesurf[_2_ Wrote:
;94178']And in practice, once you're recorded and processed using
96k/24 there doesn't seem much point in downgrading the result in the
age of cheap multi-TB drives.
Agreed. However, during a time when bitdepths and sampling rates are
sold more like commodities rather than formats, shouldn't we point out
the same file reduced within reason, to say 16-bit 48kHz, would replay
with absolutely no perceptible loss of quality experienced by the
listener?
If I were certain that were true in all cases for all listeners in all
situations, yes. But we do have evidence that there are objectively
testable reasons based on human physiology that such effects may be audible
at times for some people, etc. So I can't rule out that it matters.
FWIW When I listen to a well-made CD, or indeed LP, or the BBC iplayer, I
can often happily enjoy the music without being aware of such possible
problems. So I don't advise losing sleep over this. :-)
The snag is that all too many CDs/LPs are *not* well made. So again,
bypassing as many avoidable 'reprocessings' may be advisable as a general
strategy.
But of course, many 'high rez' files prove to have flaws. e.g. have a look
at
http://www.audiomisc.co.uk/HFN/HealthCheck/Flac.html
In particular compare the sample probability plots for the 'My Sweet Lord'
(96k/24bit flac) commercial download with the decca one of Britten's War
Requiem. The Britten sounds superb - amazingly so when you realise how long
ago it was recorded. The Harrison is perhaps less impressive in technical
terms, despite being much more recent.
So the bottom line is always that those making and processing recordings do
need to take care as well as use good kit. But once they do, one basic
sensible principle is to simply avoid any needless process because entails
a risk of some 'harm' to the final result. That doesn't guarantee a superb
result, but it does dodge some possible landmines, so makes a better result
more likely.
Jim
--
Please use the address on the audiomisc page if you wish to email me.
Electronics
http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio
http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc
http://www.audiomisc.co.uk/index.html
Linear Mode
