Why 192kHz matters?

[DoggieHowser]

This is from an MQA presentation.

 

 

 

But it does try to explain why our brain can perceive and react on sound transients in as short a time frame that even 96kHz samples cannot fully capture. 

 

Also talks about how even as we age, our perception of audio signals is not "halved" because the perception is less linear and logarithmic. 

Edited January 9, 2016 by DoggieHowser

[Guest Sime]

Bugga, I'm outa popcorn, brb......

[Kaynin]

^ lol

[Newman]

I absolutely hate these videos. There is maybe 30 secs of stuff I want to hear, per 30 mins. NOT what mp3 is and how it started, thank you. I avoid them like the plague.

 

I presume it is about timing and human perception of timing being in the microseconds. Conveniently forgetting that digital sampling is accurate to the nanosecond or less, irrespective of sampling rate.

[DoggieHowser]

I absolutely hate these videos. There is maybe 30 secs of stuff I want to hear, per 30 mins. NOT what mp3 is and how it started, thank you. I avoid them like the plague.

 

I presume it is about timing and human perception of timing being in the microseconds. Conveniently forgetting that digital sampling is accurate to the nanosecond or less, irrespective of sampling rate.

 

 

I don't get your point. Are you saying a 44.1kHz sample rate can capture such a short impulse that the brain can process?

 

I think the main difference I see in the two camps has to do with what I call steady state measurements/response vs impulse or sudden changes. 

 

The brain I think is capable of processing sudden changes that isn't always captured in a temporal "summed" measurement. 

[Martykt]

Fantastic explanation of why higher resolution files are beneficial.

The go to argument against higher resolution files is that we can only hear to 20khz therefore nothing above 44.1 khz is required.

While having frequency limits of over 20 khz is not necessary the point about capturing the music with the amount of samples that the human auditory system is capable of seems to get overlooked.

Having extra room for the required filtering seems a great point too.

I'll be interested to hear what other things Hanz Beekhuyzen has to say, he seems to have a lot of common sense grounded in scientific understanding.

[DoggieHowser]

Thanks Martin

 

Here's part 2 of the MQA presentation - which at 7 min longer - must be even more annoying to Newman

 

 

 

The MQA presentations are designed for lay people as well so I don't mind going over stuff I already know. 

[Nada]

 it does try to explain why our brain can perceive and react on sound transients in as short a time frame that even 96kHz samples cannot fully capture. 

 

 

 

That's a myth.

 

Its a persuasive concept that's become accepted as it just seems to make sense.

 

The myth goes.

 

1. humans can hear incredibly accurately in the time domain allowing a survival edge by giving us spatial sound location ie when a snake rustles near us we can immediately tell where it is and jump away so fast we only realize its a snake after we move. This spatial processing uses many cues but transient edge differentiation between the ears is one and has been estimated to have temporal threshold of 7us (0.000007seconds)

 

2. the myth then proclaims the sound frequency to encode a 7us impulse is the inverse = 142kHz ( not that it matters but the freq is actually the inverse of the wavelength of a sine wave   f=1/7x2 = 71kHz)

 

3. myth then says need to have a 192kHz sampling to embed the spatial cues and therefore deliver  a great sound with clean transients and spatial accuracy to yield that great audiophile sound with dynamic snap and superb imaging

 

Sound right huh?

 

Its all incorrect as far as I understand sampling theory science.

 

But never left the truth get in the way of  good marketing pitch.

[Nada]

Fantastic explanation of why higher resolution files are beneficial.

The go to argument against higher resolution files is that we can only hear to 20khz therefore nothing above 44.1 khz is required.

While having frequency limits of over 20 khz is not necessary the point about capturing the music with the amount of samples that the human auditory system is capable of seems to get overlooked.

Having extra room for the required filtering seems a great point too.

I'll be interested to hear what other things Hanz Beekhuyzen has to say, he seems to have a lot of common sense grounded in scientific understanding.

 

Myths are usually very catchy. Dont let them make a fool of you.

 

If you have above average hearing and can actually perceive a 16kHz tone then a sampling frquency of 34kHz  let alone 44.1kHz will do you nicely and embed temporal cues down to the accuracy of your DAC clock running in the Mega hertz range with jitter probably a thousand times better then needed to resolve to 7microseconds  ie orders of magnitude better then your temproal heating discrimination.

 

Dont let me stop you buying the latest DAC though. Maybe go with DSD so you can sample in the Mega Hz?

Edited January 9, 2016 by Nada

[davewantsmoore]

There is no doubt that the auditory system can respond at the time scales mentioned.

 

.... but

 

Hans (and others) are inferring that sampling at 192khz (or whatever, high rate) allows you to locate frequencies below 20khz more accurately in time.

 

 

Anyone (anywhere?) care to do a quick chalk and talk on how that is possible?

[Guest]

Simply cannot see how using the highest bit rate is not a good thing, particularly as digital storage is becoming so affordable. Why we are not buying hi res digital recordings on USB SSD, is beyond me. I see the CD and its transport's days coming to an end.

[Nada]

I

 

Simply cannot see how using the highest bit rate is not a good thing, particularly as digital storage is becoming so affordable. Why we are not buying hi res digital recordings on USB SSD, is beyond me. I see the CD and its transport's days coming to an end.

 

Yep Bigger is always better. Why only do kilohertz when you can run at Megahertz right?  

[Nada]

There is no doubt that the auditory system can respond at the time scales mentioned.

 

.... but

 

Hans (and others) are inferring that sampling at 192khz (or whatever, high rate) allows you to locate frequencies below 20khz more accurately in time.

 

 

Anyone (anywhere?) care to do a quick chalk and talk on how that is possible?

 

This is the visual argument (that I think is totally misleading spin):

[davewantsmoore]

I don't get your point. Are you saying a 44.1kHz sample rate can capture such a short impulse that the brain can process?

 

Words are very important here.   Depends exactly what you mean.

Edited January 9, 2016 by davewantsmoore

[Martykt]

Myths are usually very catchy. Dont let them make a fool of you.

 

If you have above average hearing and can actually perceive a 16kHz tone then a sampling frquency of 34kHz  let alone 44.1kHz will do you nicely and embed temporal cues down to the accuracy of your DAC clock running in the Mega hertz range with jitter probably a thousand times better then needed to resolve to 7microseconds  ie orders of magnitude better then your temproal heating discrimination.

 

Dont let me stop you buying the latest DAC though. Maybe go with DSD so you can sample in the Mega Hz?

As far as I can tell that the best understanding of the auditory system that Hanz Beekhuyzen is aware of.

The truth is when it comes to the auditory system and the brain there is a lot we do not understand and what we do know is probably only scratching the surface.

Music is made up of more than just a single tone or a frequency limit. It also has very intricate harmonics and micro details. The point of having more samples therefore more musical information seems a valid point.

Where exactly the limits of detail that we can hear is a question and one not easily answered, however as far as file resolution goes where we don't know exactly what those limits are it stands to reason that having more information is better than not having enough.

MP3's smaller size were chosen partly due to hard drive data limitations. As hard drive technology has increased this is no longer as much of an issue.

[Guest]

I

 

 

Yep Bigger is always better. Why only do kilohertz when you can run at Megahertz right?  

No point in making another question to defend an older now becoming redundant system. Like the analogue tape, things change, now it seems it is in favor of quality rather than hardware limitations , about time, has been a long wait.

edit, and cannot come fast enough as age catches us all

Edited January 9, 2016 by Guest

[davewantsmoore]

As far as I can tell that the best understanding of the auditory system that Hanz Beekhuyzen is aware of.

The truth is when it comes to the auditory system and the brain there is a lot we do not understand and what we do know is probably only scratching the surface.

Music is made up of more than just a single tone or a frequency limit. It also has very intricate harmonics and micro details. The point of having more samples therefore more musical information seems a valid point.

 

The information presented on the auditory system is accurate.

 

The ideas being put forward about how digital audio works aren't.     The inference that for signals < 20khz, that a higher sampling rate than 44.1khz allows these signals to be better organised in time, is wrong   (and why nobody ever demonstrates how that works)

 

Here's some what looks like 4th year EE, which covers it quickly:  http://www.ece.iastate.edu/~namrata/EE424/l1.pdf

 

 

Certainly, sampling a much higher bandwidth, makes it much, much easier to design filters....   and that IS a big deal....    but that's what oversampling is for.    It doesn't mean we need to receive our content sampled at higher rates.

[Nada]

..............

Where exactly the limits of detail that we can hear is a question and one not easily answered, however as far as file resolution goes where we don't know exactly what those limits are it stands to reason that having more information is better than not having enough.

 

Thats a very attractive idea. Reasoning relies on logic however. It does not stand to reason that bigger is better. Just because our culture attempts to brainwash us that  more is better doesn't mean its true.

 

Sampling theory isn't dependent on understanding human perception. It only needs to digitally encode an analogue signal, like the music with all its complexity, into bits and then decode it with minimal error. There will always be error but having a bigger file doesnt mean less error.

[davewantsmoore]

No point in making another question to defend an older now becoming redundant system. Like the analogue tape, things change, now it seems it is in favor of quality rather than hardware limitations , about time, has been a long wait.

edit, and cannot come fast enough as age catches us all

 

So why is it better then?  

[davewantsmoore]

however as far as file resolution goes where we don't know exactly what those limits are it stands to reason that having more information is better than not having enough.

 

They're telling you that high sampling rates are giving you "more information"   ....   is that really the case?  

Edited January 9, 2016 by davewantsmoore

[Guest]

So why is it better then?  

Watch the video's in the links provided again. Pls Dave 

I liken this to the car argument, electric is obviously going to be better but there will be some that will dogmatically defend fossil fuel ICE to the bitter end, even though it is inferior. 

[Nada]

Watch the video's in the links provided again. Pls Dave 

I liken this to the car argument, electric is obviously going to be better but there will be some that will dogmatically defend fossil fuel ICE to the bitter end, even though it is inferior. 

 

Well there's no sense in reasoning with that

[Nada]

Obviously all the sampling theory is wrong. Just look at this:

 

 

Hang on. Appearance can be deceptive and advertising often is trying to deceive us. So for anyone who wants to think for themselves instead of being brainwashed:

 

The impulse response above is for 3us.

 

3us is a wavelength of 6us and so a frequency of 167kHz

 

Even in a magical world 167kHz is just ridiculous.

 

So what sampling frequency is good enough for accurate impulse response.

 

Assume you can hear 20kHz which you know you cant.

 

Assume a second harmonic at 40kHz can superimpose to steepen the leading edge of the 20kHz and be audible which is extremely unlikely, then 82.2kHz sampling will suffice. 

 

Id suggest if anyone runs this as an experiment at home  they will find any sampling over 32kHz will be the threshold of spatial location accuracy for all subjects assuming no down-sampling  of files is done.

 

Its the down sampling which is the problem.

Edited January 9, 2016 by Nada

[LHC]

I think it helps to put something to bed otherwise there is too much cross talking. I posit that no one posting here was questioning sampling theory or how it works. The mathematics and theory behind it is not being question. Could we move on please and stop talking about sampling theory. Even the creators of MQA do not claim that anyone can hear frequencies above the usual limit of 21kHz. That is the accepted limit of our ear's sensitivity in the frequency domain.

 

The claim by MQA creators is that despite the fact we can't 'hear' those high frequencies in the frequency domain, their 'impacts' in the time domain can somehow be perceived. They maintain that in the time domain our sensory is five times more sensitive than that in the frequency domain. When that happens we are not saying one could hear frequencies of 37kHz or 38kHz (of course we can't), but what those high frequencies does is to contribute to a higher rise time, and hence sharper musical note, and this can be heard (that is their claim). 

 

Once again in plain language: when we hear a complex passage of music, our ears could only identify frequencies (or pitch if you like) up to 21kHz only, it is inherently limited. But we can hear the contributions of much higher frequencies to the rise time of the notes, and hence the clarity of the notes. 

 

I do noted Newman's point that CD sampling of 44kHz already gives us rise time in the order of nanoseconds. So the claims by MQA creators that human can hear better than that is quite extraordinary (and exciting) if proven true. 

[Martykt]

The information presented on the auditory system is accurate.

 

The ideas being put forward about how digital audio works aren't.     The inference that for signals < 20khz, that a higher sampling rate than 44.1khz allows these signals to be better organised in time, is wrong   (and why nobody ever demonstrates how that works)

 

Here's some what looks like 4th year EE, which covers it quickly:  http://www.ece.iastate.edu/~namrata/EE424/l1.pdf

 

 

Certainly, sampling a much higher bandwidth, makes it much, much easier to design filters....   and that IS a big deal....    but that's what oversampling is for.    It doesn't mean we need to receive our content sampled at higher rates.

If I have my understanding of how digital sampling works correct (not going to pretend to be anything more than a layman) a higher sample rate will be able to create a more precise sound wave. The more samples T there are the more precise X will be as it is recorded at that point in time.