What’s the point of upsampling?
Oct 30, 2024 at 3:19 AM Thread Starter Post #1 of 66

gaex86

Head-Fier
Joined
Sep 12, 2024
Posts
80
Likes
60
Location
italy
Sorry If it is an obvious question, by since I got in this world I keep on reading about upsampling, often with dac/dap or dongle.
The same I see with ripping cds or records; lot of people rip in 24 bit/dad whatever.
So my question is what’s the point of upsampling a source encoded in lower quality?
 
Oct 30, 2024 at 4:53 AM Post #2 of 66
Upsampling is a very effective placebo device.
 
Oct 30, 2024 at 5:33 AM Post #4 of 66
Upsampling is a very effective placebo device.
That is a very unscientific answer. :wink:

As I am not a scientific expert in this field, so I try to describe it in my own words, how I understand this topic.
Upsampling can be used to apply "external" (e.g. in a computer) mathematical "better" reconstruction filters when going from digital to analog when using 16bit/44.1 kHz as source. This only "makes sense" when you use a DAC, which does not apply its own internal reconstruction filters after your nice external ("better") reconstruction filters.

As already hinted, DACs usually perform an internal upsampling, then called oversampling of a 16/44.1 signal, so they can apply also better reconstruction filters internally.
If these "better" and different (internal / external) reconstruction filters can be heard or not, is very much up for debate and fills pages and pages in various threads, which is what probably will happen here. :wink:
 
Oct 30, 2024 at 8:09 AM Post #5 of 66
I have enough problems in my life to not have the need to worry about upsampling and internal/external reconstruction filters. I can't imagine my life being so problem free these kind of things would keep me awake at night or even occupy my mind.
 
Oct 30, 2024 at 8:47 AM Post #6 of 66
This only "makes sense" when you use a DAC, which does not apply its own internal reconstruction filters after your nice external ("better") reconstruction filters.
There will always have to be an internal reconstruction filter, if only the final analog part of it.
The external upsampler will - as the name suggests - output a higher sampling frequency than what is coming in. And it will digitally filter out frequencies above half the input signal's sampling frequency. But this is still a digital signal so the internal reconstruction filter of the DAC will of course still have to filter out anything above half the sampling frequency of the signal coming out of the upsampler. (Either by first doing another oversampling plus digital filter step plus the final analog filter, or just the final analog filter.)
(Some total crackpots use external upsamplers in combination with effectively defective filterless NOS DACs in which case the total system is not complying with the correct conditions for digital audio to work correctly. In those cases an external upsampler helps to improve from "unnecessary wrong result" to "unnecessary but less wrong result". In contrast: using no external upsampler and a $9,- apple dongle gives an audibly perfect result.)
 
Oct 30, 2024 at 9:15 AM Post #7 of 66
I am such a crackpot!

As said, the crackpots state that the external filters are better than the internal ones of a DAC, due to more calculating power in an external computer as internally in a DAC.
If this is audible? Some say yes and some say no. Personally, I don't care what either side is claiming.
 
Oct 30, 2024 at 10:16 AM Post #8 of 66
I am such a crackpot!
So, you use a filterless NOS DAC? Indeed then there could be an audible difference between using or not using an external upsampler. It would mitigate some of the imperfections of the filterless NOS DAC, maybe even to the point of reaching audibly transparency. But of course it is far better and easier to just use a correctly designed and functioning DAC.
 
Oct 30, 2024 at 10:25 AM Post #9 of 66
Yes, I am using a Holo Audio Cyan2.
As said far better to use the internal filters is up for debate. Easier, I agree with you. But it is a hobby and easy is not always the aim in a hobby. :)
 
Oct 30, 2024 at 10:54 AM Post #12 of 66
No worries, just stay on your path, thats totally fine. This is just nerds talking.

No but it’s cool, because I seriously think in this hobby there’s often a very little “scientific” approach and it’s nice to get a something more technical from people who know more.

I’m a, sad to say, former guitarist, with at times huge collection of fx, amps, guitars etc and I’m really familiar with all the shenanigans, miraculous overdrive, and whatever, with lot of claims and no scientific background; and this is pretty much what I see in this my new hobby.
 
Oct 30, 2024 at 12:01 PM Post #13 of 66
I guess I’m even more confused now 😂😂 but thanks for all your answers!

The simple idea:
Almost all DACs are delta sigma, and oversampling(much higher than 96 or 192kHz) is necessary as part of the design. To oversimplify, so the main concept is clear, imagine the DAC takes the digital signal and turns it into 1 and zeros but with a different code where 1 means the voltage goes up and 0 goes down. No fixed voltage value, just the given direction of something changing constantly. With very many samples per seconds, and the speed for amplitude increase/decrease being very high, the amplitude movements outside the original signal can only contain very high frequencies. It's crap noise created by that apparently crude process, but because its content is made of very high frequencies, we can just low pass the high frequencies somewhere after the music content and have excellent attenuation of that noise. Leaving us with very high fidelity.
That's the general principle and why we want oversampling, it helps keep music and crap easy to separate and at a lower cost.

In reality, it's more complicated, because true 1bit DACs are a mistake(same as to why DSD is a mistake), so almost all DACs nowadays work with a handful of effective bits?
Also, some of the oversampling(upsampling) in the DAC is done as part of a reclocking/anti jitter process. And really, in modern DAC chips, some more DSPs can be involved for various reasons and those tend to run better at a particular sample rate, which is another reason to change the sample rate as needed where it's needed.

That's for almost all DACs.


But of course, elite audiophiles won't settle for what the plebs is using(even when it's the best solution...). So you will find some true one bit DACs for a DSD playback that's as bad as the early days of DSD players, and it might also convert PCM so the signal can enjoy 1bit problems too. It turns out fine because luckily with DACs, even bad stuff tend to sound nice and measures better than the rest of the playback chain.
And of course, You can find a few R2R DACs. Some are non oversampling(NOS), which is objectively also a mistake from the old days. And that's where the genius of selling solutions to problems we created ourselves, comes into play. You can get a nice software for your computer to use the fanciest oversampling before sending the signal to a NOS DAC you probably paid more so it does less. It's a beautiful world.

I'm overly sarcastic, and I oversimplify, but it should still be mostly right.
 
Last edited:
Oct 30, 2024 at 12:59 PM Post #15 of 66
@castleofargh I general I find your assesment very accurate. But using terms like "audiophile elite and "plebs" are quite unscientific and when I recall correctly we are here in the scientific section of the forum where we should use more scientific terms. :)

 

Users who are viewing this thread

Back
Top