Originally Posted by hifisponge
My understanding is that Classe feels that apodising filters are not the Holey Grail of digital filters. They come with their own set of trade-offs.
I'm curious about how they came to the conclusion that there's insufficient merit in using them. I'm not saying their conclusion is wrong--I have no way to prove that. Was it based on actual listening experience, papers, measurements, other?
I also get the impression that they feel that they have minimized the filter ringing to a negligable level in their current design.
I think Classe has done a superb job with their DACs--and the entire SSP (bugs notwithstanding
). I take no issue with their sonic achievement.
Minimizing ringing in a DAC is different than removing (pre-)ringing from the source. Since the SSP DACs are always running at "2x" (88.2/96 kHz), and are 8x oversampling, there's no audible ringing issue left to deal with in the SSP other than that of "1x" sampled sources, 44.1/48 kHz. Upsampling them doesn't remove the ringing caused by the original antialias filter. Passing the audio thru a "perfect" DAC doesn't either. This is what the apodising filter uniquely does.
As for the pre- vs post-ringing tradeoff, natural sounds do not pre-ring. And seldom, if ever, do sounds stop as fast as they start, and they exist in reverberant spaces. Post-ringing is therefore likely to be masked, and therefore much less innocuous.