[Orpheus]

hi. it seems the subject of "upsampling" is highly controversial... and it doesn't really have much concrete scientific evidence that it actually makes for higher quality sound... or at least that's what i get from reading articles...

...but let's say it does--let's say upsampling works. ...now, what i really want to know is why upsample to 96khz!? ....the whole reason we had 48khz sampling rates in the first place was to avoid copying of CD's to DAT tapes. to change 44.1khz sampling (CD quality) to 48khz causes a loss in fidelity... and so the creators of DAT thought that this would prohibit the use of DATs for copying. .....so, wouldn't the same be true of 96khz conversion from a CD?--i would think that it would be MUCH better to "upsample" to 88.2 khz, as this is a multiple of 44.1khz. a 96khz sampling rate is directly incompatible with CDs.

can anyone explain?

thanks.

 

[JaZZ]

Good point: the 44.1/48 kHz sampling rate conversion issue.

I for one would spontaneously agree that such a conversion is lossy. But I read lots of positive and even glowing reviews about upsampling, meaning non-integer sampling-rate conversion to 96 or 192 kHz / 24 bit – last but not least from our well respected moderator Jude. Inspired from that, I have purchased a Bel Canto DAC2, and I'm very pleased with it. I don't know if it's the upsampling which makes it sound that good, but indeed there are much more details and more space on the recordings than before, with my Theta Pro basic II, and a much finer grain, as if there was a higher resolution.

An integer «upsampling» to 88.2 or 176.4 kHz is identical with the well-known and meanwhile traditional oversampling realized in common CD players. So it's definitely the non-integer conversion/interpolation combined with a word-length extension to 24 bit which makes the difference, if there is any at all. Technically there is no reason why upsampling should do things any better than oversampling, so its one of today's audio mysteries (like the sonic impacts of [digital!] cables and CD transports). Some call it a marketing hype. Myself I tend to take it seriously.

JaZZ

 

[Joe Bloggs]

According to the stuff the techies posted in the last upsampling thread, asynchronous upsampling converts jitter to noise or eliminates jitter like a relocking does. Also I wouldn't be so sure that asynchrous upsampling to 96kHz definitely causes degradation.

 

[Orpheus]

so, are you saying that if we upsampled to 88khz, which would be an easy conversion, that this is inferior to 96khz conversion, which is much more complicated and requires interpolation?

that seems pretty illogical to me.

96khz reduces jitter?--how's that? ...yes, the interpolation process does introduce noise. how is it that jitter specifically is converted to noise? i would think that if there was distortion caused by jitter in the signal, that jitter is in fact non-distiguishable. if it does convert jitter to noise, it is not only "converting" jitter, but also the rest of the sound too. so, does it convert jitter to noise?--maybe.... but at the same time i believe everything! is subjected to added noise at the same time.

besides, if you're concerned with jitter correction... there's many jitter reducing boxes out there.

so, the question is, the big names long ago set out a standard for DAT, which is very much related to our discussion here. like i said, they created the 48khz format to directly make copying of CDs impossible. so, that would imply upsampling to 96khz is also distorting sound. huge companies long ago decided on this..... isn't 96khz upsampling directly going against the ideas established long ago by these companies?

 

[AndreYew]

All upsampling requires interpolation, no matter what the ratio is.

Also, 44.1 kHz to 96 kHz can be done with "integer" upsampling, but with greater computing resources required. A 320x upsample followed by a 147x downsample will get you "integer" upsampling for 44.1 to 96 kHz.

--Andre

 

[Orpheus]

so then what about between 44.1 and 48khz?--industry claims that it is impossible to convert perfectly between the two.

 

[AndreYew]

44.1 to 48 kHz upsampling is perfectly doable with a 160x147 matrix. I have no idea what the industry says, but people like Weiss can do a very good job at SRC.

--Andre