Tír Na HiFi

BTW, for those who are up for some heavy reading about this subject - here's an AES paper from Julian Dunn (a giant in this area) which deals with exactly what might be at stake in using high res files & why http://www.nanophon.com/audio/antialia.pdf

It also neatly dovetails into the audio test files I posted looking for volunteer listeners. As stated in this paper, anti-imaging filters are used on the output of digital sampled systems to eliminate the alias images that are generated by this system. Now, in 44.1KHz systems the images start at 22.05KHz. It is very difficult to construct a filter which will allow 20Khz signals through but effectively cut off anything above 22Khz. The difficulty is that these filters have an audible effect at 20KHz & below. One of these effects is that the majority of these filters produce an attenuated pre-echo of the signal. This was thought to be inaudible but my tests & the reports from most of you (& others) show that it isn't inaudible :)

So in moving to a higher resolution we greatly relax the very steep cut-off requirement of this filter as the alias images don't occur until 48Khz (in 96Khz samplearates) & 96KHz (In 192 samplerates). So this is one reason for a possible advantage of higher samplerates.

OK, I can buy that argument. But wouldn't upsampling always be a good thing then? Or is good implementation what holds this back?

Fran wrote:OK, I can buy that argument. But wouldn't upsampling always be a good thing then? Or is good implementation what holds this back?

Yes, Fran, in theory upsampling should be a good thing but it seems that in most instances it adds it's own distortion/noise to the signal. I guess it depends on the implementation - I have noted in a lot of upsampling DACs that the upsampling masks the sound advantage of low jitter sources. When/if the upsampling can be turned off, the sound improves (with low jitter sources, anyway!) dCs, Meridian & some others seem to get it right, to a large extent, though.