To Doug Schneider,
Thanks for your answers to my previous questions. If I may, I have another.
Much has been written about the 24/176/192 files available as downloads. And the recommendation is to buy expensive sound cards that can output the high sample rates to a separate DAC. There are a number of good USB DACs that will accept 24/96 over USB. Has anyone reported as to the playback quality of playing the 24/192 files on these USB DACs, letting the software player downsample to 96kHz? And compared it with the same file played back over the 24/192 sound card and 24/192 DAC? Is there really a sonic difference?
There’s actually more than one question here, but that’s fine because they are good ones.
I’m sure that there are people who have experimented with such things but I haven’t yet, and to the best of my knowledge, neither have many other audiophile writers, because there’s not much information that's been written on the topic. Where I have heard some discussions and read some information is on the pro side of things -- in other words, the world of recording studios, recording engineers, etc. Frankly, the audiophile world is really playing catch-up right now. That’s why one of my pet projects at the moment is about sample rates and sound quality; however, it’s too early to say much. One person who is writing about things like sample rates and bit depth is our own S. Andrea Sundaram. You can find his series of articles, "Understanding Digital Music -- What Bit Depth and Sample Rate Really Mean" under the "Features" tab.
What I will say, though, is that we’ve found that upsampling and downsampling can be done extremely well, or not so well. This is true for computer audio and for disc players. For example, I really liked NAD’s C 565BEE CD player, but preferred the non-upsampled sound versus upsampling to 96kHz or 192kHz (it’s user-selectable). I thought the non-upsampled signal sounded more realistic and natural. One designer who definitely agrees with me on that subject is Charles Hansen of Ayre. Their QB-9 DAC, for example, supports 44.1, 88.2, 96, 176.4 and 192kHz sample rates, but does no upsampling whatsoever. Instead, it will only process the sample rate as it’s delivered. Preferably, that delivered file won’t be an upsampled one, but instead is the file at its "native" rate.
If one is to do upsampling or downsampling, we’ve found that integer-based conversions work best. For example, changing the same rate from 44.1kHz to 88.2kHz, or vice versa, seems to be easy to do. This is a doubling of the sample rate if you go up, simply halving it if you go down. Where trouble seems to occur is with non-integer sample-rate conversion -- in other words, going from 44.1kHz to 96kHz or 192kHz, which is what the C 565BEE does. That’s not to say that you can’t do a wonderful job of upsampling the signal, but not everyone does. The best I’ve heard has been with gear from dCS, but their equipment costs an arm and a leg. Perhaps that’s the price you have to pay to do it right -- at least for now.
Finally, to give you an answer to your question: I suspect you won’t hear much difference between 96kHz or 192kHz signals that were derived from the same file and with the same upsampler or downsampler. However, if the native rate of the file is 44.1kHz, it wouldn’t surprise me at all if it sounded better at the lower sample rate. This, though, is what must be experimented with. Hopefully we’ll be writing more about this subject soon. . . . Doug Schneider