more info on CD upsampling...please (1 Viewer)

[Rone]

I am seeing some DVD players available that offer CD upsampling from the standard 44.1khz to frequencies as high as 176.4khz. I am pretty sure this does not raise regular CDs to meet the quality of DVD-A or SACD. However, there must be some benefit (hopefully not just marketing). I am still a little confused on how this will help. The CD will still be a 16-bit source, right? What are your experinces with CD upsampling? If it will make my regular CDs sound better...I am all for it. Please shed some light on this issue. Thanks

 

[John S]

They promote it as a way to take off that "digital edge" analog audio purists are always on about. It of course cannot get any more high frequency out of a bandlimited digital source.

 

[David Judah]

It is really the same thing as oversampling which has always been used for CD playback. Since there is no information being added as John mentioned, it is presumed that the higher frequencies allow for a better analog conversion due to a better filter.

So, the fact that it upsamples isn't as important as the quality of the conversion filter to begin with, and of course, different designs yield different results--some are better than others.

DJ

 

[PaulDA]

At higher sample rates, it is easier to filter than at lower rates, so the filters are, theoretically, more effective (thus accounting for the more "analogue" sound). At least, this is what I've read in various places (I haven't experienced it myself so I can't say how well it works).

 

[Leo Kerr]

on the other hand, it's nothing new.

Anyone remember into the dark ages of the 1980s? When it was all the rage to advertise 2x, 4,x, or even as high as 12x oversampling?

 

[Philip>L]

In regard to filtering: it's not that it's easier to filter, it's that the filtering takes place in the frequency space above the human hearing range so that you're not losing any of the information in the audible range which does occur without upsampling.

Whenever a digital source is converted to analog there are artifacts that occurs at whole number fractions of the sampling rate. These artifacts lessen at the whole number increases. So by quadrupling the sampling rate you allow filtering to occur at one, two, and four times the upper limit of the audible range instead of occurring inside the audible range.

 

[PaulDA]

^^^ Thanks for the clarification. Now that you put it that way, that is more like what I remember reading.

 

[David Judah]

To expand on what Phillip is saying, since CD is sampled at 44.1, that leaves us with a bandwidth of around 22KHz per Nyquist. Our audible range is around 20kHz, so the filter has to block everything above the cutoff without affecting things in the audible range. That's a very steep slope with only around 2 KHz of leeway, and since filters aren't perfect it can cause problems that we can hear.

With upsampling or oversampling there is much more room for the filter to work with. The slope is not as steep which makes it cheaper and easier to design with less deleterious effects in the audible range.

DJ

 

[Phil Nichols]

I suspect that Denon's "AL24 Plus" treatment of CD audio is doing something like that, but so far I haven't been able to locate a white paper on their AL24 technology. I know it raises CD sound to imulate 24 bits which supposedly yields a more "analog sound" to CD's.

State of the art video engineering is also employing oversampling to more affectively filter per what David says above involving Nyquist principles.

FYI, here's a white paper on video oversampling which is "similar in principle" to audio oversampling and may help a bit with understanding the reasons for audio oversampling:

http://denon.ca/hapdfs/nsvtechnote.pdf

 

[Chu Gai]

Oversampling is a superset of upsampling which is essentially just a reclocking of your existing data to a higher rate. This makes the data cover more bits. There is no benefit per se with upsampling data that is already band limited to 20 kHz with regards to something like greater precision or smoothness. That's just not going to happen as the Nyquist theorem states that if we sample twice the bandwidth we can represent the original signal faithfully. Having more points doesn't hurt but it doesn't help either. When there is a sonic difference with upsampling it happens because the implementation has added additional information in the way of disortion which are basically non-linearities. Some people like that and don't be so surprised if that's what 'hi-end' companies do intentionally and then make up some BS about this and that and then charge you up the wazoo.

With oversampling we have the ability to increase both the accuracy and the linearity of the converters. Upsampling is done but so are other things. Prediction and noise shaping are done which alters the noise floor of the signal by moving it to higher frequencies which can then be removed by a run of the mill analog filter. Moreover, high quality reconstruction and anti-aliasing filters are then implemented in the digital domain. This makes life much easier and is probably cheaper given the rapid advances in technology.

Done properly, there should be no differnce in sound. If there is a difference, then one, or the other, or both even are 'broken'.

For those who are curious about brickwall filters and at what point they become audible, download the files over at http://www.pcabx.com/technical/low_pass/index.htm and determine it for yourself.

 

[Phil Nichols]

Chu,

I believe there is some clarification required here, IMHO.
"Prediction and noise shaping are done which alters the noise floor of the signal by moving it to higher frequencies which can then be removed by a run of the mill analog filter. Moreover, high quality reconstruction and anti-aliasing filters are then implemented in the digital domain. This makes life much easier and is probably cheaper given the rapid advances in technology.

Done properly, there should be no differnce in sound."

There should be some "difference in sound" because the higher frequencies do in fact make it possible to use cheap filters which otherwise might not be done at all to save money. Also the high quality reconstruction and anti-aliasing digital filters are then present in the design which otherwise might be left out to save money. So oversampling is (or should be) done for a reason - the reasons you give above - which if implemented will make the sound better (.....more noise free and post-processing distortion free).