bhaagensen;282186 Wrote: 
> 
> it's Easter and I seem to have time to wonder about all kinds of
> strange things, among which is subj.
> 
Happy Easter, Bjørn

bhaagensen;282186 Wrote: 
> 
> I think that according to the Nyquist theorem sampling frequencies
> above 22.05 at 44.1 results in aliasing, therefore everything above
> 22.05 is filtered in the analoge domain before sampling to avoid
> aliasing ? But why/how can one then see from the reconstructed waveform
> that clipping is present (due to e.g. loud mastering)? Modulo being a
> bit flat, it shouldn't clip as far as I understand?
> 
Actually "filtered" is not the correct term, folded would be better. If
the sample rate is 44.1 an input signal or transient with a frequency of
33 would "fold" to a copy of the signal around 33-22.05=10.95 and would
be perceived as distortion. To prevent distortion, before sampling the
signal is low-pass filtered. Because unsampling (in your CD player or
SqueezeBox) again creates high frequency components also here is
low-pass filtered.
Now clipping is something entirely different. Is Nyquist related to the
frequency domain, clipping is related to the amplitude domain. If I have
a 16 bit sample there are 65536 discrete "levels" lets say from 0db to
96db. If I input a signal of 100db the range would be to small but the
sample would be 65536 was we ran out of numbers; the signal gets quite
literary "clipped". It is than not possible to restore the signal back
at unsampling as the unsampler does not "know" that the signal was
100db
bhaagensen;282186 Wrote: 
> 
> Next question. Is compression not preferable to overly aggressive
> filtering as in above? Sure one gets a distorted waveform compared to
> the original source, on the other hand one can potentially avoid
> throwing too much stuff in the bin. If so what is an acceptable amount
> of compression? Or when is it unacceptable?   
> 
I assume you talk about audio compression or limiting (like Dolby A, B
and C or Dolby HXpro) All these systems work in the analgue domain and
where a means to deal with the rather poor S/N ratio of Tape of 50dB or
so. It would "squeeze" a signal with a range of 70dB to the available
50dB. 
Compression in the digital domain means simple removing bits from the
signal that would not be audible anyway like a soft high pitch sound
over a loud lower pitch sound. Of course it is debateble what is
audible or not. Hence the, rather heated, debate on compressed or
uncompressed. One still has to consider that you can only control the
last bit of the long trajectory the sound has made. What happened in
the studio or at mastering is out of you hands.

So use your ears, if it sounds fine for you it probably is ok :-)


-- 
th00ht

SqueezeBox v3, SqueezeCenter (7.0.1 - 17793)
Quad 303 + Two Quad Electrostats
------------------------------------------------------------------------
th00ht's Profile: http://forums.slimdevices.com/member.php?userid=15656
View this thread: http://forums.slimdevices.com/showthread.php?t=45107

_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/lists/listinfo/audiophiles

Reply via email to