Audibility of digital reconstruction filters

[Paul R]

We should probably clarify something here. When you say, "Does that mean the difference was caused by purely physical factors? Probably not.", are you saying that the difference may be caused by something that is not physical?

 

What would "not purely physical" include? Are you referring to something metaphysical, spiritual, etc.? As a reference, sound, the hearing mechanism, transducers, and associated electrical energy is all physical. Are you going outside of that domain?

 

I'm not trying to trap you, and no judgements, but it may change the tack here just a bit to know...

 

LOL! Of course I am not referring to metaphysical effects or the like, though sometimes I do think there may be a ghost in the machine. Non physical the way I used it was imprecise- pretty much I was referring to the software generating the signal.

 

Of those I think that the software makes the most difference since it is modifying the signal of course. And it s what people modify to create changes they hear. If you ignore that you wind up with just another bits is bits argument. Admittedly I should have been more precise and laid but that logic chain but it is an old and well worn argument. And leads to people making silly claims all over the place- on both sides. It will end with everyone dissatisfied and mildly disaffected. Everyone knows a part of the picture and nobody nosing the whole thing.

 

Paul

[Miska]

(b) the "smearing" referred to, although it can occur over a time long enough to perceive, only "smears" frequencies too high to hear, and

 

What is often ignored is the context where the effects happen. Studies have shown that ultrasonic frequencies alone don't have effect, but when combined with normal audio signal they do.

 

Testing audibility of these effects with artificial test signals doesn't go far, because the signals are not natural to human hearing. It needs to be tested with familiar signals brain already understands and can interpret. Of course you can train yourself to the test signals too, but it requires lot of time spent listening to the test signals and you need to be careful that you train to the correct version first.

 

People who have only ever listened music in MP3 format may feel that a lossless version sounds "wrong" to them.

 

Here's spectrogram of test I recorded with bunch of different instruments that have wideband transient nature. Recorded with my measurement mics (unfortunately not very wideband) using Focusrite Forte at 192k PCM. More the natural bandwidth is artificially restricted, more artificial it begins to sound.

 

[dc2bluelight]

LOL! Of course I am not referring to metaphysical effects or the like, though sometimes I do think there may be a ghost in the machine. Non physical the way I used it was imprecise- pretty much I was referring to the software generating the signal.

 

Of those I think that the software makes the most difference since it is modifying the signal of course. And it s what people modify to create changes they hear. If you ignore that you wind up with just another bits is bits argument. Admittedly I should have been more precise and laid but that logic chain but it is an old and well worn argument. And leads to people making silly claims all over the place- on both sides. It will end with everyone dissatisfied and mildly disaffected. Everyone knows a part of the picture and nobody nosing the whole thing.

 

Paul

 

Thanks, Paul, I appreciate that perspective. And am a little relieved.

[dc2bluelight]

What is often ignored is the context where the effects happen. Studies have shown that ultrasonic frequencies alone don't have effect, but when combined with normal audio signal they do.

 

Where could we find those studies? Thanks!

[Miska]

Where could we find those studies? Thanks!

 

Inaudible High-Frequency Sounds Affect Brain Activity: Hypersonic Effect | Journal of Neurophysiology

 

With some of the studies I think I now better understand why I find metal dome tweeters to have unbearable sound due tue +10 - +20 dB resonance around 25 - 30 kHz frequency range...

http://www.tinnitus.vcu.edu/Pages/Ultrasonic%20Hearing.pdf

Ultrasonic masker clarifies ultrasonic perception i... [Hear Res. 2003] - PubMed - NCBI

[esldude]

Inaudible High-Frequency Sounds Affect Brain Activity: Hypersonic Effect | Journal of Neurophysiology

 

With some of the studies I think I now better understand why I find metal dome tweeters to have unbearable sound due tue +10 - +20 dB resonance around 25 - 30 kHz frequency range...

http://www.tinnitus.vcu.edu/Pages/Ultrasonic%20Hearing.pdf

Ultrasonic masker clarifies ultrasonic perception i... [Hear Res. 2003] - PubMed - NCBI

 

Oohashi's results haven't been replicated except in one instance. And it was found the tweeter in use was producing IMD down in an audible range. When separate tweeters each producing a separate portion of the frequency range were used to prevent IMD, the results went away. Work was done by Kaoru Ashihara. Ashihara has done work showing that some small number of young adults hear up to 25 khz or so though the thresholds are 100 db or higher. So he has no vendetta to prove ultrasonic sound can't be heard.

 

Your other two links are for ultrasonic masking via bone conduction. Wouldn't apply to open air listening. Maybe, maybe would apply to headphones if they are connected in a way to conduct via bone.

 

Here is one of the papers for ultrasonic hearing by Ashihara showing a few can hear above 20 khz.

 

https://www.jstage.jst.go.jp/article/ast/27/1/27_1_12/_pdf

[Miska]

Your other two links are for ultrasonic masking via bone conduction. Wouldn't apply to open air listening. Maybe, maybe would apply to headphones if they are connected in a way to conduct via bone.

 

You get bone conduction through air too, of course less efficient than direct coupling, but you don't really need a lot level to have sonic impact in high end use. The topic in those studies was masking of 10 - 20 kHz tones by ultrasonic frequencies at 20 - 40 kHz range and to help deaf people gain hearing through bone conducted modulated ultrasonic. But it is related to music too, impact on music perception just wasn't the topic of those studies.

 

Ultrasonic teeth cleaner at my dentist sounds like having a 90's modem performing a handshake inside my head. And it's damn loud. And that's conduction through my teeth, while the device is not even touching (about 1-2 mm away). And it makes my gums tingle. I also know that I deeply hate sound of most metal dome tweeters, because listening those remind me of the dentist.

 

But for this transient smear of course happens precisely as side effect of removal of the ultrasonic content...

 

Loudspeakers are bad for any hearing tests. I listen primarily with headphones anyway, while working. I have time maybe once a week or once per two weeks to power up my stereo system and listen through speakers... But I can listen many hours per day with headphones.

 

Ultrasonic listening tests should be conducted with transients having unlimited bandwidth vs transients having limited bandwidth, natural sounds recorded with wide band equipment and mics like Sanken CO-100K:

SANKEN MICROPHONE CO .,LTD. | Product [ CO-100K ]

Sennheiser MKH-8020:

Sennheiser MKH 8020 - Recording Condenser Microphone Onmi-Directional - for Guitar Microphones - Acoustic;Bass;Brass;Wind Instruments;Drums & Percussion ...

or DPA 4004:

DPA Microphones :: Products

 

Of course if you are happy with RedBook, that's fine. I'm not happy even with any of the existing hi-res either. I can still easily hear difference between real life and recorded... Large part of that is due to mics, pre-amps, amps, loudspeakers, headphones and such, but one part of it is the digital side and the converter architectures. I don't have trouble hearing difference between RedBook and DSD128 and I don't have trouble measuring difference between transient performance of the two either.

 

I'm not really interested going to have billionth argument about the very same subject all over. I acknowledge that no matter what it is impossible to convince the skeptics, no matter what the evidence is.

[Don Hills]

Edit: never mind. Hadn't read to the end of the thread before posting.

[Don Hills]

Inaudible High-Frequency Sounds Affect Brain Activity: Hypersonic Effect | Journal of Neurophysiology

 

Interesting. In their tests they used musically correlated HFCs. It would be insightful to see if the same effects occurred if the HFC was not correlated - for example, a steady 26 KHz tone at similar average levels to those of the HFCs in the music.

 

With some of the studies I think I now better understand why I find metal dome tweeters to have unbearable sound due tue +10 - +20 dB resonance around 25 - 30 kHz frequency range...

 

Do you still notice this when the tweeters are fed with a bandwidth limited signal, that should (in theory) not excite the resonance?

 

http://www.tinnitus.vcu.edu/Pages/Ultrasonic%20Hearing.pdf

Ultrasonic masker clarifies ultrasonic perception i... [Hear Res. 2003] - PubMed - NCBI

 

These imply that the presence of HFCs might actually suppress (mask) the perception of components in the upper audible range. So the presence of HFCs, while not being audible in theselves, may alter the perceived tonal balance of the signal. In which case, the same effect could be achieved with bandlimited content by applying a gradual HF rolloff. Which is what you get with many tube amplifiers, and some well regarded classes of digital filters that begin rolloff early to avoid problems in the transition band.

[Miska]

Interesting. In their tests they used musically correlated HFCs. It would be insightful to see if the same effects occurred if the HFC was not correlated - for example, a steady 26 KHz tone at similar average levels to those of the HFCs in the music.

 

I would guess it wouldn't really have such effect, maybe some minor one. But based on my experiences, ultrasonics matter specifically as integral part of unrestricted natural tone, not alone. Again, think of it as part of the initial attack waveform matching a certain wavelet, rather than set of frequencies.

 

Like the recognizable difference between turning on LED light vs turning on halogen light, even if they had same color spectrum.

 

Do you still notice this when the tweeters are fed with a bandwidth limited signal, that should (in theory) not excite the resonance?

 

That resonance is excited with any input signal regardless of the spectral content... But it becomes especially bad when you for example use RedBook content with DAC that has leaky oversampling filter producing strong imagest from the top audio octave.

 

These imply that the presence of HFCs might actually suppress (mask) the perception of components in the upper audible range. So the presence of HFCs, while not being audible in theselves, may alter the perceived tonal balance of the signal. In which case, the same effect could be achieved with bandlimited content by applying a gradual HF rolloff. Which is what you get with many tube amplifiers, and some well regarded classes of digital filters that begin rolloff early to avoid problems in the transition band.

 

That's one side, but I think more important is that with unrestricted HFCs, system's step response is clean - complete opposite of brickwall bandlimited system's step response.

 

In any case, in real physical world you don't have artificially restricted bandwidth, but you are presented with all the ultrasonics the source emits. I believe it is important to capture all this as-is up to the frequency where it completely disappears in ambient background noise.

[Don Hills]

... That resonance is excited with any input signal regardless of the spectral content...

 

... so a resonance can be excited even when there is no energy in its passband? Can you show a mechanism by which this can occur?

 

But it becomes especially bad when you for example use RedBook content with DAC that has leaky oversampling filter producing strong imagest from the top audio octave. ...

 

I can understand that mechanism, although I am unconvinced that (a) it occurs at a high enough level to be significant and (b) that I would be able to hear it in the first place. But then, I tend not to notice peas under my mattress either...

[esldude]

You get bone conduction through air too, of course less efficient than direct coupling, but you don't really need a lot level to have sonic impact in high end use. The topic in those studies was masking of 10 - 20 kHz tones by ultrasonic frequencies at 20 - 40 kHz range and to help deaf people gain hearing through bone conducted modulated ultrasonic. But it is related to music too, impact on music perception just wasn't the topic of those studies.

 

Bone conduction of sound is about -60 db versus air conduction of sound to the ear. High end sound or no that is going to be masked by the air conducted signal.

 

 

I'm not really interested going to have billionth argument about the very same subject all over. I acknowledge that no matter what it is impossible to convince the skeptics, no matter what the evidence is.

 

And us skeptics for the billionth time do tire of people who say trust me I can hear the difference offered as evidence. Offer actual evidence and that is a different story. So skeptics aren't impossible to convince. I reject your faulty hypothesis about those skeptical.

[esldude]

I would guess it wouldn't really have such effect, maybe some minor one. But based on my experiences, ultrasonics matter specifically as integral part of unrestricted natural tone, not alone. Again, think of it as part of the initial attack waveform matching a certain wavelet, rather than set of frequencies.

 

Like the recognizable difference between turning on LED light vs turning on halogen light, even if they had same color spectrum.

 

 

 

That resonance is excited with any input signal regardless of the spectral content... But it becomes especially bad when you for example use RedBook content with DAC that has leaky oversampling filter producing strong imagest from the top audio octave.

 

 

 

That's one side, but I think more important is that with unrestricted HFCs, system's step response is clean - complete opposite of brickwall bandlimited system's step response.

 

In any case, in real physical world you don't have artificially restricted bandwidth, but you are presented with all the ultrasonics the source emits. I believe it is important to capture all this as-is up to the frequency where it completely disappears in ambient background noise.

 

Well Oohashi's results were that without music ultrasonics were not perceived. With music they were. This is a big hint or at least a red flag that ultrasonics might be causing IMD with lower frequencies which is what was being heard. Ashihara found exactly that. Use multiple tweeters to reduce IMD of ultrasonic reproduction and no one heard them with or without music.

 

If the bone conduction material has any relevance, it is that ultrasonic resonance of the brain can cause masking around 12-16 khz. Would seem the reverse of accurate reproduction to purposely mask what is heard. And no ultrasonics don't mask sound in that range in normal live listening because bone conduction in such conditions is -60 db below air conduction. If one wanted to mimic this masking you don't need resonance of the brain at 29 khz. You could insert your own low level noise into the 12-16 khz band to achieve the same result. You don't have to resonate the brain at ultrasonics to cause vascular excitation of the hearing mechanism.

[Superdad]

If the bone conduction material has any relevance, it is that ultrasonic resonance of the brain can cause masking around 12-16 khz. Would seem the reverse of accurate reproduction to purposely mask what is heard. And no ultrasonics don't mask sound in that range in normal live listening because bone conduction in such conditions is -60 db below air conduction. If one wanted to mimic this masking you don't need resonance of the brain at 29 khz. You could insert your own low level noise into the 12-16 khz band to achieve the same result. You don't have to resonate the brain at ultrasonics to cause vascular excitation of the hearing mechanism.

 

Hi Dennis:

I'm not feeling well tonight so I am having a hard time following this conversation properly. Can either of you comment on how tinnitus figures into this? From what I have read and heard (I have mild tinnitus myself), tinnitus is actually the ear/brain self-generating at the very frequencies where hearing loss has occurred--and for many, myself included, that is right in the 12-16kHz range.

 

I have tested myself and can still hear (up to maybe 15kHz), but it is interesting to start to think about the ringing in my head (some times worse than others) as bone conduction. I always thought of bone conduction just at very low frequencies. BTW, I bet a lot of people have some tinnitus and don't even realize it because they live in a city. Out here in the country it is eerily (no pun intended) quiet. My studio/office has a noise floor of about 36dB. At times I have wanted to stick my measurement mic in my ear to measure the frequency/amplitude of the ringing! (Just kidding; of course that would not work, though the small diameter capsule would fit snugly . )

 

Goodnight,

AJC

[Miska]

Well Oohashi's results were that without music ultrasonics were not perceived. With music they were. This is a big hint or at least a red flag that ultrasonics might be causing IMD with lower frequencies which is what was being heard.

 

No, it just demonstrates that it is not about the frequency content as such, but for example about rise time of the transient. The correlation between ultrasonic sound and base band sound is important.

 

Use multiple tweeters to reduce IMD of ultrasonic reproduction and no one heard them with or without music.

 

Use of multiple tweeters just significantly increase directivity of speakers (spoiling the test), while the radiator should be omnidirectional. Do you have a link to that study so that I could verify if they performed tests correctly?

 

If you want to test it, use something like this:

http://elac.de/en/products/4Pi_series/4Pi_Plus_2/index.php

 

Also check this one http://publications.lib.chalmers.se/records/fulltext/147171.pdf

 

If the bone conduction material has any relevance, it is that ultrasonic resonance of the brain can cause masking around 12-16 khz. Would seem the reverse of accurate reproduction to purposely mask what is heard. And no ultrasonics don't mask sound in that range in normal live listening because bone conduction in such conditions is -60 db below air conduction. If one wanted to mimic this masking you don't need resonance of the brain at 29 khz. You could insert your own low level noise into the 12-16 khz band to achieve the same result. You don't have to resonate the brain at ultrasonics to cause vascular excitation of the hearing mechanism.

 

-60 dB is way high enough to have very notable impact. And no, noise is not enough, you can search through bone conduction research and see that ultrasonics modulated with speech allow otherwise hearing impaired people to hear speech. This can and is used for hearing aids, same way it is important part of hearing experience. Bone-conducted ultrasound is heard as sub-harmonics.

 

Those masking elements are part of what you hear in real physical world, so they should exist with same correlation in the recorded and reproduced content too. I see completely no reason not to include and reproduce those, or what ever spectra instruments create, without imposing limitations.

 

And returning to original subject, I still don't know anybody who wouldn't have heard difference between different types of digital filters... I just don't quite get what you are trying to say, that design of the digital oversampling filter doesn't matter? So which of the all possible designs you think everybody in the world should be using? Which one is the one that gives perfect sound forever? I remember one colleague DSP engineer claiming to me that digital oversampling filter with 0.5 dB ripple and only 40 dB stop-band attenuation is completely fine for hifi use...

 

Sounds a bit like one local technical magazine here that ceased to do testing on CD players already in late 80's because "they are all perfect and sound the same, there cannot be audible difference". So I guess you are fine with 50€ player then?

[sandyk]

No, it just demonstrates that it is not about the frequency content as such, but for example about rise time of the transient

 

+1

[Miska]

Here is one of the papers for ultrasonic hearing by Ashihara showing a few can hear above 20 khz.

 

https://www.jstage.jst.go.jp/article/ast/27/1/27_1_12/_pdf

 

Another completely useless study:

"Hearing thresholds for pure tones from 2 kHz to 28 kHz

were measured for 15 listeners. The maximum measurable

level was more than 90 dB SPL. Although no threshold was

obtained for tones above 26 kHz, thresholds were obtained

from six listeners at 22 kHz and four listeners at 24 kHz."

 

Testing with pure sines is completely useless, because this is not about hearing tones, this is about hearing waveform shape of rising transient. This is because transients naturally alert humans about incoming danger and that's why human hearing has developed to be especially good at detecting transients, rather than steady tones. Brain is trying to automatically filter out steady tones (background noise) to detect any changes/transients.

 

So you rather need to compare band limited and band-unlimited snaps, pops and crackle where the spectrum spreads across from low frequencies to high frequencies.

 

More like this:

[ATTACH=CONFIG]11599[/ATTACH]

 

Or this:

[ATTACH=CONFIG]11600[/ATTACH]

[sandyk]

[This is because transients naturally alert humans about incoming danger and that's why human hearing has developed to be especially good at detecting transients, rather than steady tones

 

Such as the sharp "snap" of a breaking twig as an enemy approaches in the dark.

 

this is about hearing waveform shape of rising transient.

 

I have been saying the same here for around 4 years now. (admittedly without proof)

There is still much to be learned about human hearing, and this kind of thing is exactly why 16/44.1 is completely inadequate. The same applies to amplifier design where we dumb it down in many cases, by prematurely rolling off upper HF response. We need wide bandwidth and low noise to accurately portray the low level upper order harmonic structure in order to retain the original rise times of the signal. Even air conditioning running in "Quiet Mode" can significantly degrade this information.

[Don Hills]

No, it just demonstrates that it is not about the frequency content as such, but for example about rise time of the transient. ...

 

You may wish to try again to explain your point, because of course the rise time of the transient is determined by its frequency content, and its frequency content is determined by its rise time.

[Miska]

You may wish to try again to explain your point, because of course the rise time of the transient is determined by its frequency content, and its frequency content is determined by its rise time.

 

Exactly, that's what I've been saying all the time. It is completely different to have constant ultrasonic sine tone (without fundamental in audible range) or to have a transient with ultrasonic harmonics (noise-like, or with fundamental in audible range).

 

So this is not about alone sines, but the combination of N sines with content IN the audible band AND in ultrasonic with strong relation due to the steep rise...

[sdolezalek]

Testing with pure sines is completely useless, because this is not about hearing tones, this is about hearing waveform shape of rising transient. This is because transients naturally alert humans about incoming danger and that's why human hearing has developed to be especially good at detecting transients, rather than steady tones. Brain is trying to automatically filter out steady tones (background noise) to detect any changes/transients.

 

So you rather need to compare band limited and band-unlimited snaps, pops and crackle where the spectrum spreads across from low frequencies to high frequencies.

 

Miska: This makes logical sense (points in the same direction as some of the comments about pre-ringing due to filters), but it in essence asserts that we are separately sensitive to frequency and rise time through two different mechanisms. How would you conduct a test of the ability to distinguish between two levels of transient attack while holding frequency constant?

[esldude]

The eardrum would be an unusual transducer to have a limited waveform steepness at continuous signal input, but respond to steeper waveform rising transient. Most transducers work somewhat the opposite way.

[Jud]

We want to be careful not to let incomplete models limit our picture. Inharmonic attack transients may be an important part of instrumental and vocal sound. It's also a neurobiological fact that different neurons process transient versus tone information. So though we have a mathematical model that can translate perfectly between tone and time, that is not the way all sound behaves, nor the entire description of the way we hear.

[Miska]

The eardrum would be an unusual transducer to have a limited waveform steepness at continuous signal input, but respond to steeper waveform rising transient. Most transducers work somewhat the opposite way.

 

It's not the transducer, it's the detector software behind it...

[Superdad]

The eardrum would be an unusual transducer to have a limited waveform steepness at continuous signal input, but respond to steeper waveform rising transient. Most transducers work somewhat the opposite way.

 

But our ear drums have an awesomely powerful computer--evolved specifically for transient analysis--attached to them.