Here's an explanation of upsampling by Jeff Parks, from a 2005 review in Positive Feedback (http://www.positive-feedback.com/Issue17/arcamcd192.htm):

"Upsampling technology has been around for six years. At first, like many audiophiles, I was reluctant to believe the claims of more analog-like sound, better dynamics, larger soundstage, and better imaging. Who are we fooling? How can upsampling a 16-bit/44.1k signal to 24-bit/192k improve the sound, since the 24-bit/192k signal is a second-generation reconstruction of the original signal? Wouldn't there be signal degradation rather than improvement?

Nevertheless, in my experience, upsampled music almost always sounds more refined, with better bass and midrange, a larger soundstage, and upper end extension that is missing most if not all digital hash. In short, the presentation is much more musical. To put the process into simplistic terms (that is to say, ones that I understand), a 16-bit/44.1k signal is sampled 44,100 times per second. These individual samples are linked together—much like dots in a halftone picture—in order to create the audio signal. The term "16-bit" refers to how much information these samples use during this duration. With a 24/192k data stream, the signal has 192,000 dots per second instead of 44,100. This means that we have more than four times the number of dots creating our aural picture. The increased sampling rate results in a gentler up-and down-slope of the signal, which in theory results in more complete musical information. In short, the digital signal should sound more analog. The crazy thing is that this procedure—it works. Upsampling CD players and DACs beat my old reference 16/44.1k DAC/transport combination ... every time, with a presentation that allows me to get deeper into the recorded performance."

My experience is similar, and so is my level of understanding of the technology. Thoughts and comments, anyone?

It's really not that different from line-doubling. There are probably as many ways to do it as there are products on the market, and some of them are better than others.

I do like upsampled CD. That said, I would rather have a decent non-upsampling DAC than a sucky DAC with upsampling (and I do -- an Audio Alchemy DDEv3.0 -- for one of my systems).

The problem with up sampling is that the system has to guess the missing information, no dough some systems are pretty good at this but it still is a great source for errors. The more you up sample the bigger chance for error, just think of the 44.100 dots you mentioned, they have to become 192.000. That is 147.900 missing!

What you notice is as Mr. Parks mentions the more soft sound, since a lot of the "digital" sound is gone (hash he calls it). Think of it as driving down an old road of stone in a horse wagon, after a mile the size of the stones are reduced to a 1/4 of the size. This you will properly experience as a much smoother ride and give you more time to enjoy the details of the ride.

It is the same that is going on inside you're ears, the digital "bumps" are smaller. But it is the same rider only now the problem is; is the sound and sound picture the same as it was recorded and indented in the mix? It is here the roads split.

In my opinion there is no dough, if we talk ultra high-end/pro it is not ok. It will be an electronic/software engineer who decides how the sound shall be and not the microphone.

In a more normal system I don't see any problems with up sampling; it will increase the pleasure of listening to CD's but at the cost of accuracy to the source.


If I may qoute from Nagra's home page; "The digital signal path is entirely treated at the 44'100Hz frequency avoiding the possibility of introducing parasitic interference through up sampling techniques."
http://www.nagraaudio.com/highend/index.php


A coment on when upsampling started; The DAC's in mid 80's used upsampling, so it hardly started in 1999 as Mr. Parks states.

Also, one of the key factors in upsampling is that it is done in the digital domain -- if it isn't done, that "connecting of the dots" still happens in the analog domain, it's called "filtering". The theory behind upsampling is the hopes that that interpolation is done more accurately in the digital domain.In my opinion there is no dough, if we talk ultra high-end/pro it is not ok. It will be an electronic/software engineer who decides how the sound shall be and not the microphone.I don't think this is true -- they still have to know their target market, and that is music listeners.A coment on when upsampling started; The DAC's in mid 80's used upsampling, so it hardly started in 1999 as Mr. Parks states.That's oversampling, not upsampling (at least, not in the way the originally posted article uses the terms). There are several companies that use the words interchangeably, but a majority of them separate the two concepts.

That's oversampling, not upsampling (at least, not in the way the originally posted article uses the terms).
I think that they are two completely different things. If you do a search on AK, there's a thread that explains the difference somewhere... and no fair asking ME to explain it!!

Try and read this from Stanford (it is heavy reading): http://ccrma.stanford.edu/~jos/resample/resample.html

And then this from Stereophile: http://stereophile.com/asweseeit/344/

Notice all the filter limitations and remember that High-Fidelity (not Hi-Fi after DIN45500) goes up to 50kHz and down to I think 10Hz (on vinyl the Tchaikovsky’s 1812 Overture goes down to 5Hz!).
Then you see why upsampling do not work in ultra high-end /pro systems.

Nego: yes you are right, oversampling is bit related and up-sampling is sampelrate related. But Wadia states that they have been using up-sampling since late 80's: http://www.wadia.com/technology/upsampling_short.htm

Wadia states that they have been using up-sampling since late 80's: http://www.wadia.com/technology/upsampling_short.htmYeah, Wadia is one of the companies I was thinking of when I said, "there are several companies that use the words interchangeably..." They've been using oversampling, but not upsampling in the digital domain in the meaning used in the original article. It adds to the confusion rather than the education.

Thanks for the other links, will peruse later.

Upsampling works very well in ultra high-end gear, at least on mine and all the ones I've heard. It approaches the sound quality of SACD. Of interest, EMM Labs latest SACD/CD player upsamples redbook to twice the sample rate of SACD, to approximately 5.6 million times per second! Priced at $10k, in the high end market, it's really a bargin and I'm dying to get my ears on one.

"there are several companies that use the words interchangeably..." They've been using oversampling, but not upsampling in the digital domain in the meaning used in the original article. It adds to the confusion rather than the education.


Most, if not all, CD players oversample and have been for sometime as noted. All of the players that upsample also oversample.

F1nut, you are missing the point, we are talking CD's here not SACD.

Up sampling can not create a sound that is not on the CD, it can only make the wave curve smoother. It will not create sounds from 5Hz-20Hz or 22.500Hz to 50.000Hz, the sound spectrum is only from 20Hz to 22500Hz (maybe even less depending on the system).
A normal recording meant for CD is usually done in 88.2 kHz 24 bit, so this must be considered the normal ultra high-end/pro level. Up sampling from 44.1 16 bit to this sample rate will not contain the same information as the original! But it will create a risk of error compared to the source, so no - up sampling is more to be considered as a filter to smoothen the "bad" sound of CD's than creating ultra high quality audio.

And to put it in a simple way: If you spend $400-500 on an old TT, about the same on a new cartridge, put $1k in a phono stage, then spend the rest ($8.000) on vinyl records and you will get better sound than any up sampling CD-P will ever create (I do not wish to make this into a vinyl vs. digital thread, it is only an example. Digital audio will at some point be just as good as analogue audio).

A small note from wikipedia: "High quality open-reel tape frequency response can extend from 10 Hz to well above 200 kHz". - So something is missing in this up sampling scheme compared to those old tape masters lying around in the record company basements.

I'm not missing the point at all, I was talking about redbook CD's. I mearly compared upsampling to SACD.

Upsampling simply samples more points, if you will, of the original information resulting in an more accurate representation. Rather than more risk of error, there is much less.

As for your other comments, I can only presume that you've never heard good digital. I have heard high end vinyl and I'll take high end digital, thank you very much.

...EMM Labs latest SACD/CD player upsamples redbook to twice the sample rate of SACD, to approximately 5.6 million times per second...

I have heard that the EMM blows everything else away, but don't even know where to go to hear one... that and $10K ain't cheap. Here's the website: http://www.emmlabs.com/html/audio/cdsa/cdsa.html

Some review excerpts:

-----------------------------------------------
The Absolute Sound, Issue 125, August 2007
Anthony Cordesman, Golden Ear Award
"...A Superb CD and SACD player, ...it offers subtle musically natural detail, the best imaging and soundstage I've heard other than from digital masters, and excellent overall timbre without digital hardness in the upper octaves or artificial softening of the sound."

Harry Pearson, Golden Ear Award
"Quite simply, the best two-channel CD (and two-channel SACD) playback device of the many, many we've tested in Sea Cliff over the past year. ...you can listen with pleasure, and for hours and hours."

Superior Audio, EnjoyTheMusic.com
Phil Gold
April 2007 Click to read
"The CDSA SE has to be in very front rank of CD Players regardless of cost. It will do wonders for detail retrieval, imaging and accuracy complete with a fully realized bottom end and open top on the best Redbook recordings. Its failings are those of the medium itself. As to SACD, this is as good as digital gets, and far better than any Redbook CD you will ever hear."

Positive Feedback Online
Dr. Sardonicus (Rick Gardner)
Issue 30, March 2007 Click to read
"At USD $9.95K full retail, it doesn't take a genius to figure out this should soon be the hot ticket. And it doesn't matter if you don't have even one SACD. At this point in time, on planet Earth, digital music reproduction just doesn't get any better than the CDSA SE."
-----------------------------------------

If I understand this correctly, 5.6 is the rate at which some SACD masters are made. There are a handful of pro players that use it, and it's supposed to be vitually as good as vinyl. IIRC, there is a group trying to push it as the next standard, intended to obsolete SACD and DVD-A.

Any of the industry insiders here ever encounter it, or anyone hear the EMM?

As for your other comments, I can only presume that you've never heard good digital.

Well I can't say I have heard all the best, there is new stuff comming all the time but these and similar do fine with me even that they are a little old: http://www.apogeedigital.com/products/da16x.php