Upsampling / Oversampling

[Bingo]

:confused5: Often I hear about upsampling and oversampling when I read reviews about CD players. I am not a sophisticated, knowledgeable person when it comes to technical stuff...I learn a little bit here and there when I pose a probllem. As far as I now understand the process regardng CDs I believe that the 1s and 0s are 'samples' of sound and not a continuous sound wave as it was on vinyl, and so CD players try to fill in the gap with additional samples...trying to fill out the 'sample' so to speak. Now that may be absolutely incorrect, in which case some kind responder will clear it up for me...but so far that is how I believe it is. So.... this makes me wonder if there is any difference in sound from a CD player that upsamples or oversamples to 92 or 192...there is one CD player that upsampes to 782 I believe... I have heard opinions that say upsampling does not matter at all, that 92 is sufficient, that 192 is better..no different etc.. opinions go all over the place when you read commentaries and reviews. So...how does a dumb dumb like me decide how to use the upsampling feature when selecting a new CD player? I do know that I have a $179.00 Sony CD player and also a $1200.00 Cary CD 308 CD player, and I know that the Cary SOUNDS MANY TIMES BETTER THAN THE SONY...THE SAME CD PLAYED ON ONE AND THEN THE OTHER IS VERY OBVIOUSLY DIFFERENT WITH THE CARY SO FAR OUT IN FRONT IT AIN'T FUNNY. Anyway, if anyone wants to chime in I will read your responses with great enthusiasm...so have at it.

Bingo

[Feanor]

Bingo,

I not technical either but I know there is a lot of yammer about "asynchronous" versus "synchronous". The former, asynchronous, I believe, is synonymous as upsampling while the latter is just oversampling.

Crudely speaking, oversampling is reading the standard CD frequency of 44.1 kHz bits a discrete mulitiple of times; that is, 2x44.1 = 88.2; 3x44.1 = 132.3, etc. Upsampling asynchronously refers to converting the frequency to another, non-discrete multiple, say 96 or 192 kHz. Some pundits say asynchronous or up is bad for sound, but synchronous or over is OK.

(Similarly the CD bit count, 16 might be oversampled synchronously to 32, or upsampled asynchronously to 24. Presumably the former is "good" while the latter is not so good.)

[Bingo]

Hi Feanor...alias Bill Bailey...thanks for the update on the 'sampling' business. I am going to try and locate an audio engineer somehow to find out a bit more for both of us. I think I am still 'in the dark' about whether this sampling thing enhances the sound we get. I do know that my cheap Sony CD player doesn't upsample, synchronous or asynchronous (?) and it sounds pretty bad compared to the Cary that upsamples to 92, I think...92.... I wonder why some go all the way up to 7 hundred and something? Oh well.when the masters disagree, the student is free!

Bingo alias Nolan Fremin:confused5:

[Rock789]

The Moon Andromeda (http://www.simaudio.com/moonandromeda.htm)
Internal 24-bit/705.6kHz upsampling using BurrBrown DF1704 Digital Filter with 16X oversampling
(haven't auditioned this guy... probably way higher than what I want to spend)

The Mood Equinox (http://www.simaudio.com/moonequinox.htm)
BurrBrown BB1730E high-resolution 24-bit/192-kHz Digital-to-Analog Converter and 8X oversampling digital filter · Internal upsampling which achieves 24-bit/352.8kHz resolution
sounds very very nice!!! (this was part of an audition a few weeks back)

so, both these state oversampling and upsampling... how does this work?

[Feanor]

The Moon Andromeda (http://www.simaudio.com/moonandromeda.htm)
Internal 24-bit/705.6kHz upsampling using BurrBrown DF1704 Digital Filter with 16X oversampling
(haven't auditioned this guy... probably way higher than what I want to spend)

The Mood Equinox (http://www.simaudio.com/moonequinox.htm)
BurrBrown BB1730E high-resolution 24-bit/192-kHz Digital-to-Analog Converter and 8X oversampling digital filter · Internal upsampling which achieves 24-bit/352.8kHz resolution
sounds very very nice!!! (this was part of an audition a few weeks back)

so, both these state oversampling and upsampling... how does this work?

The Adromeda over- (not up-) samples 16 times, i.e. 16 x 44.1 = 705.6, (synchronous). However I note that the bit count is 24 per your info above, so maybe it upsamples the bit count, viz. 16 x 1.5 = 24, (asynchornous).

The Equinox also oversamples, in its case, 8 times, i.e. 16 x 44.1 = 335.8. However in addition to CD input, (16x44.1), it also handles hi-rez input up to 24x192. Possibly it oversamples the hi-rez, i.e. 48 x 4, or 96 x 2, = 192, although they might be calling that upsampling.

[Rock789]

Thanks Bill

[hifitommy]

that upsampling is multiplying the frequency to a higher one and then performing the filtering at the higher freq so ringing cased by the sampling isnt near the audible range, resulting in a more relaxed sound.

perhaps i am wrong but so far, nobody with technical expertise has answered here.

on the cary, the highest upsampling doesnt always sound better, but the one not the highest usually does. its been a while since i heard it at upscale audio so i dont remember what the freq was.

my sony ns500v seems like it upsamples because it sounds so much better on rbcd than any other player i've had in my system.

[Dusty Chalk]

The problem is, different companies use different definitions. The people at Wadia, for example, use oversampling and upsampling synonymously, whereas people like MSB make the distinction that oversampling is increasing the sampling frequency only, whereas upsampling is using special algorithms to increase the sampling frequency and the bit depth at the same time -- so you have to take into account that the word may have a different meaning depending on the source.

[hermanv]

Dusty is right, there are few standard definitions, certainly none at all one the marketing departments get done hyping their equipment.

Cheaper CD players may only realize 14 bits of real accuracy, step one is to get all 16 available bits, you can't do this today for $50.00. Once this is done, there are additional steps that can be taken. They are listed below in order of relative complexity (note: not necessarily in order of the improvement in sound quality).

If you increase the sample rate, i.e. process the same digital signal more times per second it becomes possible to improve the performance of the reconstruction filter and improve the resulting sound quality. A given manufacturer may or may not have taken advantage of this possibility (those marketing folks again) because just multiplying the clock is a cinch and allows you to write that on the retail box.

Next, it is also possible to sample more often and dither the least significant bit, once again if the filter is improved to take advantage of this, an improvement in sound is possible.

Lastly the most expensive and complex process is to sample more often and also increase the word size (16 to 24 bits) and to interpolate (i.e. guess intellegently what the missing bits might have been if they were there). Now you also need to again improve the filter to take full advantage of this process. People smarter than me can predict the limitations of this idea, but an extra bit or two of average resolution certainly seems likely.

All of these improvemnts are on the small side and will be audible most clearly on better systems or equipment. On many of todays more expensive players the differences in sound quality is due much more to the quality of that last analog stage than it is to improvements the digital process. It turns out that true 16 bit performance takes a damn good analog stage not to screw it up.