darrenyeats;675031 Wrote: 
> You are right, the examples tend to use a steady tone.
> 
> You say that dithering doesn't work for constantly changing signals.
> Why do you say that?
> 
> Regards,
> Darren

Because dithering works by averaging.  For instance if you had a steady
value that should be reading 9.5 but the resolution of the your ADC
could only read whole numbers, you could never get any closer than 0.5
of the correct reading.  You can only read 9 or 10.  But if you add
noise to the desired signal, and the noise is random and at least high
enough to move the reading +/1 digit, then if you took a large number
of readings of the signal that is at 9.5 it would read 9 sometimes and
10 other times due to the noise, but if you averaged out a large number
of readings, then the averaged reading would be 9.5, effectively
increasing the resolution of the system.
But if you can't make a large number of readings on the same value with
the noise added, it doesn't work.  In fact, if you can only make one
reading before the value changes due to actual changes of the input
value, then the noise just makes the reading less accurate.
Obviously, music isn't changing instantaneously, but it will rarely be
steady state enough to allow many samples of the same waveform on a
row.  In the case used as an example on the web page mention, the 1kHz
tone would have to be present for 743 milliSeconds.  That is 7 full
cycles of the 1kHz waveform.  Can we expect that a music waveform would
be that steady?  I'm sure that dithering does help to some extent on
music, but the extent would depend on the statistics of music.  Just
how dynamic is it?  I've never seen any good discussion on it.  
Another way to look at it is hinted at in the web page we are talking
about.  He talks about how making a large number of samples effectively
reduces the noise by binning the frequency into smaller bands, each
containing less noise.  But that can't apply when the signal is moving
over the full bandwidth while the averaging is happening.  You would
just average the signal along with the noise.  

Terry


-- 
TerryS
------------------------------------------------------------------------
TerryS's Profile: http://forums.slimdevices.com/member.php?userid=40835
View this thread: http://forums.slimdevices.com/showthread.php?t=89733

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

Reply via email to