Practical DSP and Oversampling

Limitations of analog anti-aliasing and anti-imaging filters motivate a practical digital filtering approach in which high rates are used for sampling and reconstruction. To avoid excessive computational or storage cost associated with high rates, the sampling rate is changed by downsampling the discrete-time signal prior to processing/storage and then upsampling the signal prior to conversion back to continuous time.

[hllwmn]

Good lecture!
In the audio press (not recording, but playback, as in CDs and D/A converters), I've seen the digital (OS) filter also -- perhaps erroneously -- referred to as a "reconstruction filter".
The argument:  Because the digital filter ADDS interpolated data to the original, the filter "reconstructs" the signal.
This is confusing (maybe the audio press folks are confused?!) ... I thought the REAL reconstruction was done in the D/A converter ... and that signal is then filtered in the analog domain (reconstr. filter) before output.
Perhaps the term FILTER used in oversampling filter -- aka digital filter -- is somewhat incorrect -- it makes filtering easier for the ANALOG FILTER, but doesn't really low/high/bandpass the pure digital-domain data.
Is this correct?

[Roscovanul]

I love you for this fine explanation :*

[gustavlicht]

Question regarding the anti-image filter - could the band pass compensatino part be implemented in DSP, before the DAC, and the anti-imaging filter then reduced to a low pass filter with high suppression of the frequency range where the image is formed?

[王柯罗]

Thanks for your video, it helps a lot!