Re:A Phoenix record store owner set the audiophile world on ...
by PaulDavisThe1st on Friday August 05, 2022 @03:53PM
(#32361154)
> with a 96 kHz sampling rate, you don't need such a brick wall filter.
This is incorrect. See my comment above/below.
A sharply cutting off analog filter is expensive to produce. It has multiple stages to create the multiple poles. High precision resistors and capacitors have to be used to get all those circuit stages to line up. The filter will have phase distortion.
That's the basis of "supersampling": sampling at a higher rate with a simpler filter with less of a cutoff, then completing the job with a digital filter to get to the target sample rate.
This can be done in reverse, in reproducion. Take, say, a 48 KHz signal, and digitally interpolate it to a higher sample rate like 96 KHz. That is fed to the DAC. The filter after the DAC then doesn't need such a steep cutoff after 20 KHz. A greater bit depth can be used; like 16 bit samples interpolated to 24 bit at a higher rate, fed to a 24 bit DAC.
The digital filter or interpolator doesn't care about accurate resistors, capacitors or drift in component values over time or due to heat; it does the same thing with the same data every time.
Digital filters can look at future values also. The state of an analog filter is determined by only the current and past values of the signal; but digital signal processing can delay the signal a little bit and look at a "box" around the current value. I think that is key to preserving phase relationships.