I've been out for a few days, so not meaning to drag that other thread out, I'll just post my responses in this thread. Out of ignorance, I will probably not participate, though I wish I could get some of the companies who perpetuate "upsampling" (Musical Fidelity, et al) as different from oversampling to participate. In the current issue of Stereophile, Anthony Michaelson of Musical Fidelity chides Stereophile for using upsampling and oversampling interchangeably, stressing that they
are different. Sure wish the guys at Stereophile understood this well enough to explain it. Quote:
Originally posted by nec
Thank you for trying to explain this to such an idiot like myself... |
Sorry about the snideness on my part, was still on "momentum" from arguing with Ricky. You know "quote your sources!" Sincerely, I do apologize. I obviously know just enough about this to be dangerous -- know just enough to think I know more than I do. All I really do know is that they do sound different to me. And upsampling sounds better to me.
I am serious about learning more about DSP -- would the books that you mentioned in that other thread be good books to get? I was also thinking about getting the CSound book. Quote:
Originally posted by Joe Bloggs
1. What have you got to say on the topic of what's the difference between over- and upsampling? Seeing as that article shows that oversampling already includes interpolating intermediate values... |
You're kidding, right? I thought I'd stated this, but I'll state it one last time -- it is not my understanding that oversampling includes the kind of interpolation that exists in upsampling. It certainly couldn't have interpolated to 24 bits back in the early 80's, when it was first introduced. I have to believe that that article is incorrect in implying that all oversampling DAC's perform interpolation. I
do understand Andre correctly in that they are put through a filter (read: same thing) after oversampling, but it was not my understanding that it was part of the same process. My understanding of the process -- previously -- was that they just repeated the last value. It has since been made clear that they usually just introduce zeros.
Maybe the big difference is the context. Oversampling was always done in a DAC -- it had to be part of the process that after the signal was converted to the analog domain it was put through an analog filter. None of these things were ever taken out of context and expected to sound adequate in an isolated context, the designer was only interested in the overall -- D-to-A -- process. Upsamplers can act as standalone products, so their output has to be an improvement on the input in a wider variety of circumstances. Some upsamplers are D-to-D, some are daughter boards, just as two examples.
But again, and I'll stress this, I'm guessing. This is based entirely on a bunch of advertising literature -- I have yet to find any technical discussion from any of the manufacturers -- and magazine articles. Quote:
Uh, bear with me--so are the OS chips in upsampling DACs set to oversample as well? E.g. In the specs for the ART DI/O I saw 128x oversampling as well as upsampling to 96kHz... |
This is another example of using the term "upsampling" in its classic DSP sense, not in the current hi-fi equipment sense. In this sense, it is no different than oversampling.
Again, I blame the manufacturers for using a term that had a long standing meaning in a different sense. But I stand by my belief that they are different. Quote:
Originally posted by MirandaX
Strictly speaking, CD audio is not perfect. First, even though 44.1kHz sampling is sufficient to perfectly represent any <20kHz signal given an ideal DAC, there is the 16-bit quantization issue. For loud recordings, it doesn't tend to matter, but for wider range classical (e.g. Mahler's 1st symphony, or the 4th movement of Mahler's 2nd symphony), there is a perceptible difference in resolution as signals get quieter. |
Although I agree with Mr. X (
) in his assessment, there is an additional usefulness in going 24-bit -- not just near the noise floor. I also feel that (true) 24-bit recording allows for more separation between instruments, providing better imaging and soundstage as well. Upsampling (to 24-bit
with interpolation, not just dithering) should provide
some improvement in this arena as well (although this is a glaring case of "you can't create what isn't there").