What is the point of upsampling?

[Eminar]

This is the corresponding EQ filter:

View attachment 62461

And this is the phase shift caused by that filter:

View attachment 62462

Cool. Group delay?
I'm not imagining that I've heard smearing in many forms over many years.
Nothing ever free

 

[QMuse]

More gain more delay

Sure, more gain more GD, larger Q more GD. But we are acctually speaking of very modest phase/GD changes. Here is how it looks for a filter with +6dB gain (which is a lot) qith Q=8 (which is also quite high):

As you can see phase shift barely reached 22deg, which is nothing. But that amplitude correction would certainly very audibly improve frequency dip if applied correctly.

 

[Eminar]

Yep
That's true, and this is not aimed at you, but no it's not on personal preference.
under educated people use the vocabulary they are stuck with until they decide to improve it
Some opinions are worth more than others due to experience and effort

 

[PaulD]

Serious question do you actually think that the sound of that filter is what 2 inch tape sounded like
Next question have you heard 2inch?
Would you really think that your plugin is world class, to be used by pros the world over with repeatable results, like 2 inch tapeI said just another dicky bit of software that I'll get updated soon forgotten about
Yeah you're claim it the same but it's fucking not and you know it
Don't be disingenuous'

I have heard LOTS of 2" tape, also 1", 1/2" and 1/4", with Dolby A, Dolby SR, etc, and at 7.5, 15 and 30 IPS speeds, over many years (over a decade).

I have not tested the UA tape plugin against a tape machine, but I have tested the UA 1073 preamp plugin against a real 1073 preamp. Level-matched blind tested in a studio with other engineers. We could not tell the difference between the UA plugin and the real Neve 1073 preamp across all of the settings. We did notice that the settings matched across all of the controls. This test was done in 2019. We were all VERY impressed with the emulation, so much so that one of the engineers has sold his 1073 preamps.

I have heard the UA tape machine plugin and it sounds like tape to me (except for dropouts). Judging by the 1073 experience, I have no reason to doubt they they have had the same attention to detail.

I think these days with sophisticated modelling, the onus its now on the naysayers to prove that a well engineered plugin is NOT representative of a good sample of the modelled device.

 

[QMuse]

Cool. Group delay?
I'm not imagining that I've heard smearing in many forms over many years.
Nothing ever free

I posted GD in my first post - blue line on the GD graph. As you can see that filter improved phase and GD of the speaker quite significantly.

 

[March Audio]

Can they be made leak-proof and phase accurate?

Leak proof???? No idea what that is.

Phase accurate yes

 

[Eminar]

Sure, more gain more GD, larger Q more GD. But we are acctually speaking of very modest phase/GD changes. Here is how it looks for a filter with +6dB gain (which is a lot) qith Q=8 (which is also qutie high):

View attachment 62463

As you can see phase shift barely reached 22deg, which is nothing. But that amplitude correction would certainly very audibly improve frequency dip if applied correctly.[/QUOT

Faircall
I was never suggesting the benefits did not outweigh the cost in most circumstances
Used sparingly

 

[March Audio]

Including AD-DA?

Yes.

Try the test that @Blumlein 88 made with 8 AD /DA loops and see if you can hear the difference.

 

[Eminar]

I posted GD in my first post - blue line on the GD graph. As you can see that filter improved phase and GD of the speaker quite significantly.

Ok. Been drinking for a while and I'm on my phone I can no longer see the graphs properly
Was this eq applied to programme material or pink noise?

 

[Eminar]

Does that mean no then, a digital low-pass crossover filter doesn't need to be complemented by low-pass analogue filter?

Correct

 

[QMuse]

Ok. Been drinking for a while and I'm on my phone I can no longer see the graphs properly
Was this eq applied to programme material or pink noise?

That is a room EQ filter that is applied by to anything being played. That particular graph was measured using sine sweep. You can't measure phase/GD using pink noise or music.

 

[Eminar]

I have heard LOTS of 2" tape, also 1", 1/2" and 1/4", with Dolby A, Dolby SR, etc, and at 7.5, 15 and 30 IPS speeds, over many years (over a decade).

I have not tested the UA tape plugin against a tape machine, but I have tested the UA 1073 preamp plugin against a real 1073 preamp. Level-matched blind tested in a studio with other engineers. We could not tell the difference between the UA plugin and the real Neve 2073 preamp across all of the settings. We did notice that the settings matched across all of the controls. This test was done in 2019. We were all VERY impressed with the emulation, so much so that one of the engineers has sold his 1073 preamps.

I have heard the UA tape machine plugin and it sounds like tape to me (except for dropouts). Judging by the 1073 experience, I have no reason to doubt they they have had the same attention to detail.

I think these days with sophisticated modelling, the onus its now on the naysayers to prove that a well engineered plugin is NOT representative of a good sample of the modelled device.

Ok I can accept that

 

[Blumlein 88]

If @Koeitje is interested here is the 8th generation test.

https://www.audiosciencereview.com/...-choose-the-8th-generation-digital-copy.6827/

And an earlier version with different equipment used.

https://www.audiosciencereview.com/...dac-loop-vs-the-original-can-you-hear-it.448/

 

[Eminar]

That is a room EQ filter that is applied by to anything being played. That particular graph was measured using sine sweep. You can't measure phase/GD using pink noise nor music.

Do you think it's more likely to be audible with complex program material. I get that the math dont lie

 

[QMuse]

Do you think it's more likely to be audible with complex program material

No. Improvements in phase/GD response can be heard in a room with less reflections and with simple (jazz alike) high quality recordings.

 

[Julf]

I asked again do you have practical experience the sound of 2 inch tape

I already responded "yes, of course". I am old enough to have worked with ancient technology.

 

[Julf]

That's true, and this is not aimed at you, but no it's not on personal preference.

So where is the proof and evidence?

Some opinions are worth more than others due to experience and effort

Evidence-based facts are always worth more than opinions.

 

[QMuse]

I am old enough to have worked with ancient technology.

 

[LTig]

Nope. Try this. Mic/Pre/Amp/Speaker. Switch to Mike/Pre/A-D=D-A/Amp/Speaker. Match SPL's
I cant say one is better than the other, but they are different. Transparent?

I don't think that I have mics at hand which are good enough for this purpose. If an anlog recording is too bad to prove transparency of a digial chain so be it.

However we can be very sure that an analog recording is less transparent than a digital recording. Distorion of all kinds, compression and noise are far worse for analog regardless of the quality of the analog recorder, even when compared to very cheap ADC chip.

 

[witwald]

There is no value in upsampling. Best case is that you get what you put in there. Worst case is some loss.

Is it really true that there is no value in upsampling? For example, a CD signal sampled at 44.10kHz has a nominal maximum usable frequency of 22.05kHz. If we want a bandwidth out to 20kHz, then the analog filter will need to roll off at a rate of 680dB/octave or so (96dB of attenuation over 2.05kHz). That's very steep. Hence, wouldn't the analog filter produce a lot of phase shift within the audio band?

On the other hand, upsampling by 4x would require an analog filter that covers the range from 20kHz to 88.2kHz while producing the same final 96dB of attenuation. That's only about 45dB/octave, more than an order of magnitude less than the original filter with no oversampling. This can permit the use of reconstruction filters that produce less phase shift in the audio band.

Of course, on a sine wave test, the oversampled waveform will be more or less identical to the original waveform, subject to some variation due to the arithmetic that has been performed on it in the digital domain by the digital filtering that's required.