Phil Leigh wrote:
> The "numerically zero" part works only in the digital domain. Once in
> analogue world we have the noise floor of the source and soundcard to
> contend with... plus the tiny drifting of the source sample clock...

I thought we were in the digital domain, doing analysis after you have
run the signals through a ADC.

> The software does a very sophisticated job of aligning the two signals
> in time, minimising any false "nulls".

Which is usually done with an autocorrelation. There are other approaches.


> The software tells you in dB how far down in the mix any difference
> is.

Even 70 dB down is not 'audible' to non-canines.


> Listening to the difference - boosted by 60dB so you can actually hear
> it! - is interesting. Sometimes its just white noise - sometimes you
> can hear remnants of the music + noise...

Interesting, clearly if its white/pink/whatever noise, its not important.

Any guess as to why you can sometimes here music in the noise?
Can you hear the melody? or just some impact/transients?

I'm surprised that you say "tiny,tiny difference:
1) FLAC vs bitrate limiting via LAME to 256"

Does this vary with different types of source? I would expect things
like a jazz trio, or Tony Rice on guitar solo would be more than tiny
tiny. Unless the software fixes the phase errors.

> It would be useful if others could comfirm or deny any of my findings
> so far...

Interesting, this might make me connect up my M-Audio Delta 1010 and do
some testing. Its no longer high end, but its a decent ADC

-- 
Pat Farrell
http://www.pfarrell.com/

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

Reply via email to