MQA creator Bob Stuart answers questions.

[Ron Party]

Should the question be amended: is there anyone who has listened to a nuclear explosion and remains alive to talk about it?

 

[Blumlein 88]

Perfect analogy.


Actually, just from the first principles, it ought to be just as wide. I think that the varying parameter is not just the frequency range of individual hearing though. I believe it is music "processing capacity", more accurately described by the information throughput, expressed in bits per second.

Regarding the olympic-class dead-lifters: it turns out that they possess very high percentage of fast-twitch muscle fibers (https://www.active.com/fitness/articles/fast-twitch-vs-slow-twitch-muscles), compared to general population. So, it is not about just strength, it is also about speed.

It all starts with a genetic predisposition. Then some of the people exhibiting this predisposition get selected for specialized training, usually at a young age. Those who achieve high enough level by mid-school start competing, getting up to a very serious level of competition by high school. Those who often win the competitions (or auditions, or job interviews), get to do it professionally, further sharpening the skill.


My other hobby is volleyball videography. I started taping my daughter's games, long ago, at 1080p 30 fps. Those videos didn't quite look like live games.

When I upped it to 4K 30 fps, there was a huge qualitative change: the resolution became high enough to capture facial expressions, while still keeping the whole court in view. We could now see live, emotional, people playing on the court! To some degree, we could now feel what they feel.

When I upped it to 4K 60 fps, there was yet another qualitative change: the movements became visually smooth. When played frame-by-frame, we could now see details of individual techniques - some of heretofore mysterious secrets of volleyball mastery became plainly obvious.

For instance, it is an accepted notion that no human can produce a gross motor reaction earlier than in 750 milliseconds after the onset of stimuli warranting such reaction. But, lo and behold, the video plainly shows this particular young beach volleyball player managing to move and receive the ball just 400 milliseconds after it was spiked. Over and over again.

Or another player - and this brings us closer to the topic of "unreasonable" resolutions - executes a half-wave with the palm of her hand, significantly changing the trajectory of the ball she is spiking, in the course of just three frames, that is, in just 50 milliseconds!

I got very excited when I saw it for the first time on a frame-by-frame replay, and immediately brought it to the attention of my daughter. "I already know" - she said - "I can see this feint when I'm blocking her".

How's that for a reality check? In real time, I could barely see the gross outline of the technique being used, no matter how closely to the court I stood. My daughter could perceive the fine temporal structure of what's going on, and this in turn helped her to acquire a reputation of a player not to be messed with on a block.


Very true. Perhaps one of the reasons I'm able to perceive some of those pulses is because most of the time I'm listening to professional studio monitors in the near field. Or to a chamber live music: indeed, grand piano sounds quite different from 6 meters away vs 18 meters away.

Once again you opt to go to a unrelated field and make reckless equivalencies with wild interpretations at whatever level needed to support your idea that ultrasonics and your peculiar idea of transients is important. I'm not surprised increased video resolution and frame rate are visibly better. Video doesn't reach our level of perceptual resolution yet. Then twist it some more to make it fit (and it still doesn't). Training can let you hear and interpret things you otherwise couldn't. No training will let you hear 160,000 hz when most people don't make 20 khz. This is no ought to from first principles in the sense you portray it.

 

[pkane]

Should the question be amended: is there anyone who has listened to a nuclear explosion and remains alive to talk about it?

Don’t know if anyone has, but thanks to Sergei we know of a few rats that survived a 220dB blast in the ear, albeit with their hearing permanently damaged by this ‘listening’ exercise.

 

[SIY]

Don’t know if anyone has, but thanks to Sergei we know of a few rats that survived a 220dB blast in the ear, albeit with their hearing permanently damaged by this ‘listening’ exercise.

Would you PLEASE stop mumbling and speak up?

 

[pkane]

Would you PLEASE stop mumbling and speak up?

Just turn up the volume a bit above +220dB

 

[LTig]

There are two major uses of splines in audio time domain processing I know of:
(A) Extrapolating between samples during upsampling.

AFAIK upsampling is done by adding zero samples and then lowpass filtering the result. This is described in Richard G Lyons Understanding Digital Signal Processing, a very good book. According to him intrapolating is not a good idea.

EDIT: corrected extrapolating to intrapolating (thanks to @SIY to point it)

 

[LTig]

Krunok said:
"You still haven't grasped the simple fact that microphones used in 99.999999999999999999999999999999999999999999999999999999999% of recordings simply don't capture anything any where near 96 kHz bandwidth. Specialised mics are required to do this. Your speakers can't reproduce up to 96kHz bandwidth. You can't hear 96kHz bandwidth ."

So, your trained listeners are from Krypton, special mikes were used to record up to 96kHz and also special speakers that can reproduce all that bandwidth. Did I get it right?

The simple fact is that a physical event that only lasts 1 microsecond, such as a nuclear bomb explosion, can be heard, because it puts in motion a chain of physical phenomena, which eventually create a sensation of sound. You don't need a hearing frequency range up to 1,000,000 Hz to hear a nuclear explosion, or a sound of lightning, or a supersonic shockwave, or a sharp transient generated by cymbal/xylophone/gamelan.

Your argument has no value. Even if those sharp transients exist:

• it makes no sense to try to capture them by an ADC if the microphones cannot capture them in the first place
• it makes no sense to try to reproduce them by a DAC if the speakers cannot reproduce them
• it makes not much sense to try to reproduce them fully in an audio chain because they are inaudible.

What humans actually hear are those frequencies of the transient which fall into the audible range. This can be a lot, just go outside during a thunderstorm. This also let you know how the air works as low pass filter.

 

[SIY]

Between samples is interpolating, not extrapolating. Pedantic, I admit.

 

[Blumlein 88]

I think this thread has examples of exo-strapolating.

 

[LTig]

Between samples is interpolating, not extrapolating. Pedantic, I admit.

Of course, and you are not pedantic to correct this.

 

[Sergei]

Why are you comparing a nuclear bomb explosion to music replay in the home?

Because relevant parts of the two models, expressed in the language of Physics, are same.

The hearing range of humans has been firmly established over decades of scientific research. There is a tiny fraction of the populous that can hear into the mid 20 kHz range and the SPL needs to be huge to do so. There is no credible scientific evidence that contradicts this.

Never argued otherwise.

You still haven't grasped the fact that the definition of transient does *NOT* mean there is ultrasonic acoustic content, or that you are perceiving it if there is. This point is easily demonstrable. Record one of these "transients" such as a xylophone, high pass filter it above 30kHz and replay. Let's see what you can hear. You will hear nothing.

You can find papers on tons of experiments involving transients. Turns out, they are more perceptually important than their energy proportion would suggest. For instance, removing the transients (~5% of energy) reduced the intelligibility of speech in the presence of noise by ~12% in one of such experiments.

Mixing and mastering engineers know that attenuating the transients too much will make it more difficult for the listeners to detect onsets of quickly arriving music phrases. For some music genres, it is unimportant. For others, it is.

Why do you propose to high pass above the 30 KHz? 20 KHz is more appropriate.

If you think you have heard normal music recordings that have bandwidth up to 96kHz then you are wrong. The mics didn't record the sound up to that frequency and your speakers didn't replay even if it were there.

Frequency components up to 80 KHz on the FFT graph of a 192/24 live music recording? I've seen them. Only those are not necessarily the fingerprints of ultrasonic sinusoids (which I agree humans can't hear). They could be also indications of high-frequency noise and pulses.

A pulse - let's say one sample 2/3 maximum amplitude up on 192/24 PCM, in the middle of otherwise silent for 2 seconds sound fragment, is easily heard. If you physically disconnect, or acoustically block, all the transducers, with the exception of the tweeter, the pulse is still heard - at least on my gear.

Sorry but you have diverted this thread from its subject with unsupported waffle. Whatever evidence that is supplied that contradicts your assertions you ignore and keep on waffling.

My take - sorry, you've been straight yet impolite with me, so I'll be that way too - so, my take is that understanding Physics is not your strength. Because of that, it is very difficult for me to precisely explain to you what I mean.

Some of the members here did present evidence that contradicted my prior beliefs. I accepted that, publicly acknowledged it, and adjusted my beliefs. You did not present such new to me evidence.

I respect the products you are selling. Nicely designed, beautifully finished, proper gear. By extension, I respect you. So, while I can't explain the physics to you, I will try to illustrate my point through an imperfect yet perhaps more understandable analogy.

If you sell an item via online store, the process of customer paying you may only take 250 milliseconds, from the time the customer pressed the button on the final order form, to the time that the order was physically captured by the e-commerce system - an imperceptibly short period of time for most humans. It is the later physical events that the order triggered, which make you perceive that the money paid is now yours, and enjoy whatever goods and services you decide to spend it on.

Analogously, a pulse transfers mechanical momentum to the cochlea, in as short time as 5 microseconds. It is the later events - in this case the vibration of basilar membrane, later amplified by the outer hair cells, and later yet sensed by the inner hair cells, that produce the sensation of sound. The whole process takes hundreds of microseconds, let's say 500 microseconds in a specific case.

If we translate this back into the time scale of the online ordering process, the characteristic time of you noticing that the money arrived to your account would be 0.250 x ~100 = 25 seconds. This is in the ballpark: perhaps you set up an email or a mobile text message alert to notify you of such events. The point is: at some stage of the process, a step that transferred the important value was very short, virtually imperceptible. Yet its consequences are very much perceptible.

Continuing the analogy. Imagine now that your business is thriving (maybe it already is, don't know), and you are getting thousands of orders per day. You, being now a big boss and super-busy man, can't be bothered with emails or text messages about individual sales. You are still interested in knowing when the next 1,000 cables are sold, or the next 100 amps, or - let's make it more concrete - when the running revenue count crosses another $20,000.

So, you set up an alert, which only notifies you when a threshold of the integral over time of your revenue stream (== increment of the total sum since the previous alert) has crossed another $20,000. The integral includes all the sales - both small and large. The time of arrival of such notification could be most affected by a small number of large sales, by a large number of small sales, or by a mix of those.

The hearing system works in a similar way. Instead of integrating money over time, it integrates sound pressure. When a threshold is crossed, an auditory nerve fiber spike occurs, alerting the "big boss", the brain, about an event worthy of the brain's attention.

Deepening the analogy. Imagine now that the business has grown so large that you are now operating in 24 countries, which corresponds to the approximate number of human hearing system critical bands. In some of the countries (critical bands) there are lots of "music", that is, sales. You set up high thresholds for the revenue crossings for them, and don't pay much attention to them at the moment (the hair cells responsible for those critical bands are saturated).

However, you are concerned that in some other countries, into which you want your products to break, sales are not so good. Let's say the sales in France are not taking off at the rate you expected. You set the revenue crossings alert at a lower level for France. This is analogous to listening attentively to sounds in that critical band. Ability to do that is a part of critical listening skills.

Combination of the integration mechanism and different thresholds for different critical bands explain nicely why human hearing system can perceive sounds which appear to be insignificant. As it relates to pulses (rare purchases): they are faithfully integrated along with the sinusoids (purchases from steady, repetitive buyers).

Even if a particular pulse (rare purchase), by itself, is below the threshold of hearing, it will be still integrated, and this will affect the time of arrival of the sound perceptual event (the crossing of the revenue increment threshold). Not recording the pulses (rare purchases) has an effect of underestimating the running integral of the sound pressure (underreporting the running count of the total revenue) in a particular critical band (country).

Was this easier to understand?

 

[Peter Leyenaar]

It's all making perfect sense now.

 

[LTig]

[..]A pulse - let's say one sample 2/3 maximum amplitude up on 192/24 PCM, in the middle of otherwise silent for 2 seconds sound fragment, is easily heard. If you physically disconnect, or acoustically block, all the transducers, with the exception of the tweeter, the pulse is still heard - at least on my gear.[..]

1. Such a pulse is an artificial signal which is not bandlimited to 96 kHz. A DAC would add some ringing at 96 kHz (the amount depends on the characteristics of the reconstruction filter) and probably aliasing components as well.
2. This pulse contains all frequencies from 0.5 Hz up until the maximum depending on the risetime (the extreme of this is the dirac pulse which contains all frequencies up into infinity) with equal amplitude.
3. Therefore you can hear such a pulse, even when using only a tweeter, as long as the tweeter emits frequencies below 20 kHz.

 

[SIY]

Unfortunately, the basics of Fourier analysis and Shannon-Nyquist haven't sunk in, despite repeated (ad nauseum) attempts to point this out- or they were always there and this is just trolling. Either way, I'm not seeing the point of feeding it. It has nothing cogent to add to our collective knowledge.

 

[March Audio]

Very true. Perhaps one of the reasons I'm able to perceive some of those pulses is because most of the time I'm listening to professional studio monitors in the near field. Or to a chamber live music: indeed, grand piano sounds quite different from 6 meters away vs 18 meters away.

No. You are not perceiving ultrasonic signals. See above. Pulses are inappropriate test signals because they contain all frequencies.

All this is firmly the product of your lack of understanding of the subject and a vivid imagination.

 

[March Audio]

Because relevant parts of the two models, expressed in the language of Physics, are same.

No.

You are being utterly idiotic or trolling to compare the two. There clearly is no relation in the forces involved.

 

[March Audio]

Never argued otherwise.
?

No. That's exactly what you are doing. You are arguing that ultrasonic signals can be heard.

Even when your speakers can't reproduce them.

You must understand this simple concept so I can only conclude you are trolling

 

[March Audio]

Why do you propose to high pass above the 30 KHz? 20 KHz is more appropriate.


Frequency components up to 80 KHz on the FFT graph of a 192/24 live music recording? I've seen them. Only those are not necessarily the fingerprints of ultrasonic sinusoids (which I agree humans can't hear). They could be also indications of high-frequency noise and pulses.

A

So that only signals in the ultrasonic range are present. Ref the answers above about pulses containing all frequencies. You can now turn up the volume as much as you like to make it easier to hear that ultrasonic content.

You will hear nothing.


I have seen then too. At very low levels, maybe more than - 80dB down. The music has little energy at those frequencies, the mics have little response at those frequencies and your speakers can't reproduce those frequencies and you can't hear those frequencies.

I'm not going to respond to the rest of your waffling post and irrelevant post, it's a total waste of time. You are clearly trolling or so stuck in your own misinformed dogma that it is a futile exercise to engage with you further.

A scientific forum is clearly not the place for you.

 

[March Audio]

Unfortunately, the basics of Fourier analysis and Shannon-Nyquist haven't sunk in, despite repeated (ad nauseum) attempts to point this out- or they were always there and this is just trolling.

Either way, I'm not seeing the point of feeding it. It has nothing cogent to add to our collective knowledge.

I put Sergei on ignore almost immediately, I could see very quickly where this was going to go. As the thread continued I became curious so took him off ignore. A mistake.

There is such a lack of understanding of the basics that surround the discussed issues that anyone engaging with him is truly wasting their time and energy.

 

[Sergei]

Video doesn't reach our level of perceptual resolution yet.

Well, according to "experts", writing in 2013, 1080p did reach our level of perceptual resolution, and 4K was pointless:
https://www.nbcnews.com/technology/...phones-surpass-limits-human-vision-2d11691618

And yet I'm composing this message in front of a 5K monitor. 4K videos look gorgeous on it. 1080p videos look inferior. Shall I believe the "experts"? Or my own perception, and basic math?

Before investing in 4K gear, I calculated the linear resolution of 1080p at the distance from typical location of my camera to end of the opposite side of a volleyball court. I don't remember exact number now, but it was about a centimeter.

A face of a player serving a ball from the opposite side, captured at 1080p, was an emotionless mask - quite scary upon magnification actually. Capturing at 4K fixed this uncanny valley issue.

I'm realizing now that most of the classical symphonies, and quite a bit of music that could be called symphonic rock, sounded uncanny to me on CD, and that's why I almost stopped listening to these genres for a while. SACD fixed that.