Is Hi-Res Audio better because filter is outside audible range?

[antcollinet]

So you haven't heard it then?
Please spare me the inevitable reply, I know exactly what it will be.

You didn't spare me your obvious reply, so I'll not spare you.

The engineering and psychoacoustics tells us there is no meaningful difference to be heard. I don't need to listen to know that. But I've also heard electrostatic speakers - and I'm sure they were not playing anything other than high res based on the number of salespeople in the room, and the cost of the kit playing it. I was unimpressed. Amongst the weaker of the speakers at the show IMO.

See also (for my standard and obvious link that I'm sure you've already read - but this isn't really for your benefit):

24/192 Music Downloads are Very Silly Indeed

people.xiph.org

Me too.

And yet somehow you think you can hear stuff that is beyond the ability of even the proverbial teenager with perfect (but still human) hearing.

 

[manisandher]

Have you ever heard blu-ray audio through electrostatic speakers?

Sold my Quad electrostatics a while ago. Do Stax X9000 headphones driven by a Kevin Gilmore Grounded Grid energizer count? (If not, what about a 109dB/W@1m fully active horn system?)

I have a Qobuz subscription. I choose 16/44.1 whenever it's available. Well recorded/mastered 16/44.1 albums sound absolutely phenomenal... through both my headphones and speakers .

 

[welwynnick]

And yet somehow you think you can hear stuff that is beyond the ability of even the proverbial teenager with perfect (but still human) hearing.

Absolutely not, for two reasons.
Firstly, my post was about ways that anti-imaging filters degraded audio fidelity INSIDE the audible range, not outside.
Secondly, my hearing is probably the same as most on this forum - far worse than when I was a teenager.

When I was in my 20's I could hear 21.5kHz. That doesn't mean I can hear 21.5kHz now, but it does mean there are lots of other people out there right now who CAN hear 21.5kHz. All of this isn't for me - my golden ears days are long gone. It's about establishing the right standards for audibility and record / replay formats. In 1982, 16/44.1 was the best we could do, but we moved on from that level of technology very quickly, and now there's no excuse for not having a replay chain that can exceed anyone's hearing.

I usually try to be pragmatic about most things in life, but in this instance I hold my hands up - completely guilty - I'm being insufferably dogmatic about it instead.

Does an audio system actually NEED to exceed anyone's hearing? No. Do I need a car that can do 200mph? No. Can I handle the performance? No. Can I drive over 70mph? No. Do I still want a car than can do 200mph? You bet!

 

[pjug]

Firstly, my post was about ways that anti-imaging filters degraded audio fidelity INSIDE the audible range, not outside.

I don't really have a problem with your argument, but in your post #23 your 44.1 KHz filter example is minimum phase. You should have shown a linear phase filter like others presented that show good phase beyond 20kHz. I hope you would not choose to listen to a minimum phase filter instead of a linear phase filter and then complain that the phase response shifts in the audible band.

 

[welwynnick]

I don't really have a problem with your argument, but in your post #23 your 44.1 kHz filter example is minimum phase. You should have shown a linear phase filter like others presented that show good phase beyond 20kHz. I hope you would not choose to listen to a minimum phase filter instead of a linear phase filter and then complain that the phase response shifts in the audible band.

I take your point, but I wasn't trying to show any particular filter as being good or bad.
Every filter is a compromise, and we can select our favourite balance of compromises, like it's a matter of taste, rather than right or wrong.
My point was more that if 44.1kHz was adequate, we would just use the one and only correct solution, and we wouldn't need a choice of filters at all.

 

[manisandher]

Firstly, my post was about ways that anti-imaging filters degraded audio fidelity INSIDE the audible range, not outside.

As I explained in post #8, the effects of well-designed anti-aliasing/anti-imaging filters WITHIN the audible range sit at around -200dB. Don't take my word for it - it's easy to verify.

 

[pjug]

My point was more that if 44.1kHz was adequate, we would just use the one and only correct solution, and we wouldn't need a choice of filters at all.

I am sure whatever you would want would be fine with me. But I doubt you could ever get everyone to agree on a single solution. Some say 192KHz is not enough with standard filters.

 

[Mart68]

I take your point, but I wasn't trying to show any particular filter as being good or bad.
Every filter is a compromise, and we can select our favourite balance of compromises, like it's a matter of taste, rather than right or wrong.
My point was more that if 44.1kHz was adequate, we would just use the one and only correct solution, and we wouldn't need a choice of filters at all.

I have a two DACs here which have six filter options, they all sound the same to me so I'm not sure we do need a choice of filters. Or at least I don't.

There's no question that 44.1KHz is 'adequate', the fact that proponents of higher sampling rates for playback have to immediately resort to edge case situations to defend that position should be indicative of that.

 

[welwynnick]

My interest in phase response came from seeing what phase correction could do for the step response of speakers. Good electrostatics were always my gold standard for transient response (though rubbish at everything else). I get that many people don't care about transient response - that's fine - but it does matter to me. Some Buckhardt, Neumann, Genelec and Dynaudio moving coil DSP active speakers have started using phase correction, and got similar step response to electrostatics. Then I saw that Trinnov and Audiolense could do similar things for passive speakers, and the penny dropped. Good time domain response comes from a combination of linear frequency response and linear phase response. Not just one nor the other, it's got to be both.

Those are mostly mid-range effects, say 200 to 2kHz. I wondered if a similar approach could be applied to DACs at high frequencies. There aren't many phase measurements around, but Soundstagenetwork have started doing phase measurements of (amps and) DACs recently, and they showed phase and impulse results for each filter type. I was looking for minimal roll-off in the pass band, effective attenuation in the stop band, linear amplitude response, linear frequency response and linear phase response. All the tests at 96 and 192kHz were fine, but at 44.1kHz, the results would always be compromised in one measure or another, depending on which filter was selected.

For example, here's the phase response of one my DACs, which uses the outdated Wolfson WM8741.
The top line is the frequency response (left Y axis), the bottom line is phase (right Y axis). This uses the DAC's built-in linear phase filter. I eyeball that as roughly +/- 3* around zero, near perfect.

This test was different though. This seemed to have ideal filter characteristic AND good phase response. Can you confirm that there was good attenuation in the stop band? Can you confirm what sample rate you used for this test? When I first saw it I thought it must be at 96 or 192kHz.

But this is what all anti-imaging filters should really look like:
(Stereophile measurements of Ferrum Wandla D/A.)
Full attenuation before Nyquist.

I never paid much attention to Ferrum Wandla, but both the Stereophile and SoundStage tests were very good. Again, with the HQ Guass filter selected, both the filter response and the phase response were ideal even at 44.1kHz . I didn't think it was possible to avoid those compromises, but this gives me a lot more confidence in 44 / 48kHz.

 

[danadam]

My point was more that if 44.1kHz was adequate, we would just use the one and only correct solution, and we wouldn't need a choice of filters at all.

AFAICT we don't need it. Yes, it is offered but only because there is demand. Whether the demand is for rational reasons is arguable.

As to filter correctness, here's how I see it.

When capturing the music, the filters on the ADC side has to be fairly steep in order to avoid aliasing. I don't know if it is common to have selectable linear/minimum phase filter type there, but let's say it is. Here are two 1 kHz square waves "A" and "B". "A" is band-limited with linear-phase filter, "B" with minimum-phase filter and both are sampled at 44.1 kHz:

The job of a DAC, in my opinion, is to reproduce that band limited signal exactly.

Let's look first at "A" and upsampple it 4x with linear phase and minimum phase filters:

Now let's look at "B" and do the same:

So to me only linear phase in a DAC is "correct". But I don't stress about it because they sound the same to me.

 

[MRC01]

... I was looking for minimal roll-off in the pass band, effective attenuation in the stop band, linear amplitude response, linear frequency response and linear phase response. All the tests at 96 and 192kHz were fine, but at 44.1kHz, the results would always be compromised in one measure or another, depending on which filter was selected.

This test [WM8741 filter response graphed above] was different though. This seemed to have ideal filter characteristic AND good phase response. Can you confirm that there was good attenuation in the stop band? Can you confirm what sample rate you used for this test? When I first saw it I thought it must be at 96 or 192kHz.

The WM8741 filter that I graphed above was at 44.1 kHz sampling and has a stop-band of 24.1 kHz (-112 dB). So it does stretch the transition band past Nyquist, but most DACs do this when sampling @ 44.1 kHz. While incorrect from an engineering perspective, it should not introduce any audible problems.

Some of the newer DACs (such as the Topping E70) have flat amplitude & phase response to 20 kHz and a proper stopband at Nyquist (22.05 kHz), when sampling @ 44.1 kHz. The compromise they make is a slight amount of passband attenuation and ripple - with the E70 filter #5, that is -0.5 dB @ 20 kHz and 0.1 dB respectively.

At any rate, it is technically correct to say that filters at 44.1 kHz are compromised. They are close to perfect and the compromises should be audibly transparent. But increasing the sample rate even slightly to 48 kHz should obviate the need for these compromises.

 

[antcollinet]

]

Firstly, my post was about ways that anti-imaging filters degraded audio fidelity INSIDE the audible range, not outside.

Not audibly so if the correct filter is selected. And certainly not inside the 15kHz range you say you now have.

Secondly, my hearing is probably the same as most on this forum - far worse than when I was a teenager.

Indeed - mine gives out at around 13.5khz.

 

[OHtaru]

AFAICT we don't need it. Yes, it is offered but only because there is demand. Whether the demand is for rational reasons is arguable.

As to filter correctness, here's how I see it.

When capturing the music, the filters on the ADC side has to be fairly steep in order to avoid aliasing. I don't know if it is common to have selectable linear/minimum phase filter type there, but let's say it is. Here are two 1 kHz square waves "A" and "B". "A" is band-limited with linear-phase filter, "B" with minimum-phase filter and both are sampled at 44.1 kHz:

View attachment 447669

The job of a DAC, in my opinion, is to reproduce that band limited signal exactly.

Let's look first at "A" and upsampple it 4x with linear phase and minimum phase filters:

View attachment 447670

Now let's look at "B" and do the same:

View attachment 447671

So to me only linear phase in a DAC is "correct". But I don't stress about it because they sound the same to me.

That's why I use linear phase only. Linear phase filter changes the signal once but minimum phase filter changes the signal every time. I mean if the signal is already band-limited, band-limit it again with linear phase filter makes no change, but minimum phase still makes change. It will be a disaster if filtering an audio file 100 times with minimum phase (just out of curiosity), all the transients sound like shaking a piece of metal plate. Not mention it increases inter-sample peaks. If music producer prefers the flavor of minimum phase and does it in recording process, the minimum phase effect will be baked into the music. Even if playing it with linear phase the 'minimum phase flavor' will be there.

 

[Y~BNA]

Isn't a large part of this discussion totally moot? The part where hearing above 20kHz is claimed, and sampling rates above 48kHz said to better represent HF content? And the parts where the phaselinearity of LR4 is neglected, as if 360° presents a time domain problem (it does not)?

1.
Some people's claims notwithstanding, NO ONE, and that is not a single member of humanity since science exists, has been proven able to even hear 20kHz.
When you're a healthy youngster with exceptional hearing, you still won't hear anything at 20kHz.

The science is clear and unequivocal about it. 20kHz is a theoretical upper limit, taken liberally to make sure outlyers will still fall within, with room to spare.

Anyone arguing differently carries the burden of presenting evidence, because science has unsuccesfully sought said evidence for close to 120yrs.

2.
If sounds above 20kHz make a difference to what we perceive (which is pretty much certain, because those will have harmonics at frequencies below 20kHz), they will do so at the time and place of the recording and will thus be captured in the (hopefully, if done by a sensible engineer) limited 20kHz bandwidth of that recording. No need to 'filter them back in'.

3.
Real professionals with real knowledge have filtered HiResDVD's (and HiDefCD's, etc. etc. that sounded better and measured better than regular CD's) back to 16/44.1, to test if they still sounded better.

They DID, proving that 16/44.1 one was not their limiting factor. Rather the extra attention and care that was put into those recordings had to be the reasonable explanation as to how they could sound better, as they still did so after downsampling to 16/44.1.

Professional audio engineers do seem to agree upsampling has uses, i.e. when treating a signal in the digital domain without entering the 96dB DNR of 16bit, for example.

4.
These days, double blind listening tests with control group confirm, time and again, every well designed component in the audio chain is SOTA, completely transparent to the human ear, even at 16yrs of age.

EXCEPT FOR LOUDSPEAKERS IN ROOMS.

Wouldn't we do better to focus our debate, our thoughts, our efforts and our moneys on that?

Just sayin..

 

[welwynnick]

Isn't this discussion totally moot?
1.
Some people's claims notwithstanding, NO ONE, and that is not a single member of humanity since science exists, has been proven able to even hear 20kHz.
When you're a healthy youngster with exceptional hearing, you still won't hear anything at 20kHz.
The science is clear and unequivocal about it. 20kHz is a theoretical upper limit, taken liberally to make sure outlyers will still fall within, with room to spare.
Anyone arguing differently carries the burden of presenting evidence, because science has unsuccesfully sought evidence for close to 120yrs.

I used to hear 21.5kHz, but I don't have any evidence, just a witness, Andy the workshop engineer at KJ WestOne, London. He was driving the signal generator.

The evidence is out there, but you have clearly decided not to look for it or read it or believe it. Lost cause.

Merry Christmas.

 

[Speedskater]

a) Some young people may be able to hear well above 20 kHz under some test conditions.

b) No one can hear above 20 kHz while listening to music under reasonable conditions.

c) Some digital circuit designs may have problems with signals in the 20 kHz neighborhood.

 

[Y~BNA]

I used to hear 21.5kHz, but I don't have any evidence, just a witness, Andy the workshop engineer at KJ WestOne, London. He was driving the signal generator.

The evidence is out there, but you have clearly decided not to look for it or read it or believe it. Lost cause.

Merry Christmas.

Hi there Welwynnick, a very merry Xmas 2u2!

Your attempt at framing me to be the hopeless denier is a reversal of how this works.

You performed a double sighted test, in which both researchers (you and your witness) were hoping for evidence they then imagined they found.
Mind you, i am not denying you found it, but i am denying you've proven anything.
If you had, that would be braking news indeed, under normal circumstances at normal listening levels, as relevant in ASR.

Your signal generator was hooked to an acoustic transducer, so someone could test whether they would hear a sound @21,5kHz.
This transducer, by the sheer nature of transducers, thus also produced harmonics @10,75kHz, 5,375kHz, summing effects, etc.
There is not a shred of evidence, or even any probable cause, to assume those are not the reason you perceived a sound.
Actually, rather the opposite, it is very probable that is what produced your perception.
Perhaps 50dB down, but near the centre of human's most sensitive frequency range.

Were you measuring with a mic at the same time and place your ears were listening?
Then you undoubtedly noticed the mic registered above effects, and a couple more.
What is your point exactly? You heard something? Of course you did.

There's a reason test setups are meticulously designed, thought through and described by scientists, before 'evidence' is presented as possibly valid. And even then, peer reviewing will put everything to the test again. That is why your phone, your car and your amplifier function. Science is far from flawless, but its methodology works.

Cheers!

 

[Audiofire]

Science is far from flawless, but its methodology works.

The limit for air conduction set at 20 kHz also depends on amplitude, although there is a bone conduction path that can be perceived in a similar way in the brain.

 

[Y~BNA]

Sure there's more ways to perceive sound than just through the ear-canal, more media than air, etc.

Sure there's all kinds of specific circumstances and exceptions, so it all depends. But here, we're on an audio website.

Sound propelled in air @1Bar (around sea level) @room temperature @average humidity level, can be perceived by the human ear (in already exceptional cases) up to somewhere close to 19 or 20kHz.

Sound that makes your skull vibrate and is hence perceived by the aural neurons in your ear may be a harmonic of frequencies higher than 20kHz, but it is also one hundred percent distortion of any audio signal. Because, making bone damped by blood and tissue vibrate above 20kHz is highly improbable, unless we're talking lots of energy, as in 150dB SPL, where your life will be in danger. So it will still be only harmonics of the 50kHz sound that reach the aural neurons, not an original 50k sound, and forget about audio signals altogether. It also will have nothing to do with how it sounded in air. Just like Erykah Badu singing under water will sound nothing like Erykah Badu.

What is it with this obsession to prove something that is entirely marginal and totally irrelevant to music or any reproduction thereof?

 

[antcollinet]

So it will still be only harmonics of the 50kHz

You keep talking about harmoncs *lower* than the fundamental. Subharmonics right?

How are you thinking these are created. They are not created in speakers as far as I know - nor in the air, nor from normal room responses.


Or are you thinking of intermodulation distortion - requiring more than one frequency to be present. But if so, these are not harmonics.