Cochlear nonlinearities

[John Dyson]

The fact (it is a fact) that ultrasonics can produce audible effects indirectly (even by some nonlinear effects in the human hearing apparatus itself), my guess is that the levels where the ultrasonics can produce much of a change are not created by the typical signal levels of >20kHz material on most normal recordings (outside of a lab or inside of studio.)

 

Just looked at the above 20kHz on a DolbyA encoded selection from a full 192k/24bit copy of Simon & Garfunkel '1966 Parsley' album.  On a spectogram with a 100dB range on Audacity, there were two tones - a very visible one at 28kHz and a weaker one just above 43kHz.  All of the 'splats' (musical or distortion created) start from below 20kHz.  A few of the splats maintain significant strength all the way up to the 43+kHz range, which also seems to be the final rolloff of the hiss.  (This would correlate with the DolbyA HW.)

 

Interesting observation --  the level of each splat at 20kHz seems to be less than the level of the tone at 28.8kHz.

About the 'splats' -- from what I see, the fact that they extend WAY beyond 20kHz, all the way up to 43kHz seems that it is likely to be electronically generated.  Most microphones from early '60s are more than just a few dB down above 20kHz, and ribbons can be -10dB or more at 20kHz.  Even condensers of the timeframe were sometimes not very flat to 20kHz -- looking at a U47 freq response graph - the one that I saw stops at 15kHz, and was on its way down.  Even if fully excited, I doubt that microphones like that could maintain an output representing the input much beyond the middle 20kHz range, even with decades of attenuation.  The more up-to-date U87Ai is -6dB to -10dB at 20kHz.  There is not going to be much at 30kHz at all.

 

Since the music material comes from microphones, and the typical microphone of the '60s age is a condenser likely -10 to -20dB just above 20kHz, unless it is a ribbon where it is likely down a lot more -- those splats don't appear to be generated by the microphone, unless it is the microphone itself distorting because of a huge impulse.  The splats appear to be electronic from somewhere, some place in time.  Physically complex mechanical devices seem to rolloff pretty quickly above a certain point -- but SOMETIMES might peak at a weird resonance frequency very decoupled from the normal frequency repsonse.

 

Now, I am not claiming that all apparent splats are just distortion, and sometimes there IS linear, sonically produced energy up there, but must be of a very very high level to hear (even for the magical ones who can hear 21kHz), or to directly detect distortion effects.  Of course, analog electronics especially can change behavior below 20kHz, when there is strong material above 20kHz applied to it.  Very strong signals in the distortion region above 20kHz can even shift the bias of some circuits (it is actually the result of severe IMD.)  The levels on the recording appear to be FAR FAR below what people can hear at 20kHz (or even 18kHz.) 

 

Effects from true ultrasonics (those in the air, not electronically) being audible are not quite the same as being necessary to produce any music effects (unless those effects might be to physically damage someone, for example.)

 

John

 

 

.

[John Dyson]

26 minutes ago, sandyk said:

John

 The attached is from Barry Diament's new Kay Sa album. This is genuine music content to >50kHz

Alex

 

If the material above 15k is less than -40dB (or even greater), you aren't going to hear anything except distortion effects.  Energy outside of the audible range (which requires very high level at 20kHz, even for people who can hear it) is more of a burden than anything else.

The existance of non-audible high frequency energy is meaningless when it isn't harmful, and can be a burden.

 

Think about this -- marginally audible sounds at 20kHz (for the fortunate few) are going to produce more distortion effects than anything else for us 'normals'.

 

John

[John Dyson]

18 minutes ago, sandyk said:

John

 Many of us don't give a damn about the theoretical side , we just prefer 24/192 or more recently DSD.

I shouldn't even be able to hear clear differences between even 128kbps .aac and , 16/44.1 and 24 /192 due to my age and hearing damage,  but I  can, and I can demonstrate this under non sighted conditions.

 I think that you  also know that by now too. Have my reports to you about the material you have provided been far off the mark so far ?

It's now up to the Scientists to find out why humans are able to hear these differences.

 

Kind Regards

Alex

I never argue about what people prefer, I  argue about what can be heard.

John

[John Dyson]

32 minutes ago, JanRSmit said:

Is this proven, or your assumption.?

Many years ago, audax had a tweeter going to over 40khz. We did a test plating the tweeter out of sight in a room and played a 30khz tone. When people came in the room for meetings, after a short while they became restless, sort of disturbed, but no clue as what caused it. Then after stopping playing, the restless behaviour dissapeared. I also remember Goldmund , just started then, referred to research by military dealing with human audibility of impulses, concluding that the audibility threshold is well above 20khz. 

Are you claiming that all senses of vibration are encompassed in what is heard?  That kind of 40kHz audibllity might not be through hearing, but another sense of vibration.   I can detect some kinds of vibration from senses other than hearing.

 

Even if the 'ears' detect that vibration, people weren't saying that they 'heard' the vibration, but rather they became 'restless'.  Okay, some kinds of so-called music makes me restless, but obviously not the same mechanism as the 40kHz vibration does.

 

John

[John Dyson]

8 hours ago, JanRSmit said:

How we as human beings sense vibrations is definitely not limited to our hearing system. And how we interpret the vibrations we sense is not only music of noise. It also triggers emotions and body movements like tapping with your feet, etc. 

 

 

 

There are also DEFINITELY nonlinear mechanisms that can even be used to ultrasonically direct vocal messages to individuals from a long distance -- however, we aren't typically talking about such levels when listening to music from normal transducers.  Normal level ultrasonics aren't enough to make this situation happen -- it requires relatively high levels to drive the hearing system into deep nonlinearity.  Also, such mechanisms would not have exceptionally linear transfer functions.

 

[John Dyson]

12 hours ago, fas42 said:

As someone who has been there many, many times ... it doesn't require the "magic" presence of particular frequencies to make playback deliver visceral impact, the immersive, involving quality that live music possesses... rather, it's the lack of certain types of distortion; the term "effortless" is often used in this context, and that's precisely what's needed.

Yes -- that is the main thing.  Distortion is the big bugaboo...  Excess bandwidth opens up to more IMD reaching back into the audible range.  Almost all normal recordings (not ALL, but allmost all) into the 1990s'  have been touched by analog noise reduction systems.  Such systems are a nice, very effective source of distortion.  Such distortion is usually not very measurable by simple (even multiple) test tones.

The war that audio perfectionists should be fighting (if they are true perfectionists) is DISTORTION and good frequency response up to about 20kHz (plus or minus.)  This also includes good transient response.    Distortions are what makes the difference when there is an excess frequency response and material 'up there'.  Of course -- we who know -- a lot of the 'material' above 20kHz in actual, distributed recordings is indeed NR distortion (that is, for the material into the 1990s.)  Whatever 'audio' there is above 20kHz just adds to the distortion in the audible range given real-world analog hardware.

PS: part of my reason for extreme sensitivity against distortion is that I have heard/seen how much damage is done by at least some of the old NR systems.  It is amazing how bad.  If the distortion could be measured as easily as normal HD or NORMAL IMD -- it would never have been tolerated.

 

John

[John Dyson]

On 7/12/2019 at 7:15 PM, fas42 said:

 

Except I worry about distortion on the replay side only; what exists in the recording itself I don't touch - because a) it's too much work, b) every recording will be different, c) 'simple' manipulation most likely will only introduce more audible problems, d) and most important of all, a very high standard of replay SQ allows one to hear through all the technical issues of the recording - I still find it remarkable how effective the ear/brain is at filtering out stuff that it knows doesn't belong, given half a chance.

Oh man -- you are right about the 'replay side' being hyper problematical.  That IS the bad side of NR systems -- the record (encoding) side  on the Dolby A/SR systems is actually a bit more 'pure'.  I keep on doing a successive improvement -- just did one on the DHNRDS yesterday (will probably not be in the first actual release), where I made a slight mistake in an improvement -- already still blowing away the HW version.

 

Bottom line:  I agree with your sentiments -- but also have to say -- we (the public in general -- not audiophiles per se) have been accepting quite a bit of distortion in our listening material.  It has been 'hidden' by the fact that such distortion isn't measured by simple test tones -- and the NR systems can appear to be pretty clean with those kinds of tests.  It is kind of like the 'TIM' of the olden days.

 

I have been listening to better decoded material -- so beautiful -- but makes me really feel bad that the normally available copies are either decoded with distortion or feral DolbyA.  The ONLY reason why I can hear the more clean version (DHNRDS IS NOT PERFECT) is that I have some software that really digs into the signal.  I have some 'Olivia Newton John' recordings that are -- oh so clean and beautiful.  Basically, I have NEVER heard some of the songs sound so very clean.  (For the curious, a few examples can be provided -- I do a mass decode based on the latest experimental decoder every night.)

 

The compander with complex interactions (like A and SR) doesn't make accurate decoding very easy.  DBX is easy to decode, but has other defects.  Doesn't seem like there were any perfect answers!!!

 

John

[John Dyson]

3 hours ago, JanRSmit said:

Ginny that recordings side is considered more pure. I visites a mixing and mastering specialist and whilst very impressive, i also leaned that what we get as music has no pureness whatsoever, bar an isolated case. At least in the pop music space. 

Man oh man -- from what I have seen/heard -- you are SOOOOO correct.  For a REAL trip -- listen to ABBA DolbyA material.  As a hint -- the variability in the consumer versions partially comes from the 'interesting' quality issues with the more raw versions.

I was just in the midst of a discussion with someone on another forum -- I spent several hours trying to make a copy of 'I Have A Dream'  sound clean and also similar to a consumer release -- I had several very different versions for a target, all sounded different (very very different.)   On one song (SuperTrouper), proper handling can be eye-opening clean -- not just the audio blobs/fuzzed up chorus, but ABBA is still SOOO variable.

 

On the other hand, with just a little finesse -- it wasn't difficult at all to produce very nice sounding Carpenters, Simon&Garfunkel or other such material.   (I am NOT claiming that Carpenters is pristine -- they didn't seem to produce pristine material either, but they are in better shape than some other material.)

 

Getting back to your comment -- I have just been thinking of the material that I have -- the Bread album is really good, nice, clean.  One or two ABBA, a few Carpenters are 'okay',    The Cars album was amazingly good sounding, Olivia Newton John -- variable, but generally REALLY good.  (Some of the ONJ material that often sounds pretty blah or even kind of bad on original releases, sounds beautiful when re-decoded with up-to-date technology.)  *When I speak of this or that -- I am speaking of DolbyA encoded material after being decoded*.  So, basically talking about a 1/2 step closer to what was recorded -- not including the 'mastering' or 'remastering' that is so common nowadays.

 

There is definitely a lot of pop garbage over the years.  Of course, nowadays, they take the old substandard technical quality pop material, and effectively use FM broadcast processing that totally eliminates the notion of dynamic range.  The distributors seem to want to destroy material that needs TLC, the old material does not need further destruction!!!

 

John