Jawed
1000+ Head-Fier
- Joined
- Dec 7, 2014
- Posts
- 1,124
- Likes
- 583
I have a theory that the timing errors produced by non-Nyquist DACs (or, if you prefer, their filtering) can be demonstrated with a simple difference method.Neither can I, but the audible effect is not due to us hearing 22khz content directly from a frequency domain perspective (humans cannot hear above generally 20khz). But rather the time domain information that then provides.
As stated earlier, whilst we cannot hear frequency domain information above 20khz, there is evidence showing we can hear time domain differences as low as 10uS which would require a frequency domain bandwidth of about 100khz to describe.
https://phys.org/news/2013-02-human-fourier-uncertainty-principle.html
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3663869/
Using a WAV file of white noise, create a second file that reverses the samples from the original file.
Pass both files through the DAC (or filter) and then reverse the resulting samples of the second file. In theory this second file, after having been reversed twice, will have the same samples as the original file.
I'm guessing it won't, because of time domain errors introduced by the filter. Performing a difference of the two files will presumably produce a file that has some kind of noise.
I guess the noise will vary in its character depending upon the nature of the DAC/filter...