Thanks for pointing this misleading data out, its another serious problem with the article. If you are going to do comparisons like this the frequency bands have to be perceptually relevant.

Quote from: Amir link=msg=0 date=

At 3.5 Khz and eyeballing it, the noise floor of the content is around -25 db spl and the peak 90 db.  So the total range is 90+25=115.  How did you get the 70 number?

Nice job of cherry-picking from misleading data.

The noise spectral density is said to have been measured using a 1 Hz bandwidth, but in fact the ear hears in critical bands which are about 1000 times wider around 3.5 KHz. The well  known fallacy of measuring the perceptual qualities of music or ambient noise in narrow, constant frequency bands is being exploited.

Good morning Arny .

Quote from: Arnold B. Krueger on 2014-11-28 02:46:30

Quote from: Amir link=msg=0 date=

At 3.5 Khz and eyeballing it, the noise floor of the content is around -25 db spl and the peak 90 db.  So the total range is 90+25=115.  How did you get the 70 number?

Nice job of cherry-picking from misleading data.

Boy that is a grumpy way of starting a reply .  I have now seen this phrasing used multiple times to dismiss out of hand various data point.  It must be part of the parlance of this forum.  It is another form of FUD so I hope we don't keep using the tactic when we are having an educated and deep technical discussion.  More below.

Quote

The noise spectral density is said to have been measured using a 1 Hz bandwidth, but in fact the ear hears in critical bands which are about 1000 times wider around 3.5 KHz. The well  known fallacy of measuring the perceptual qualities of music or ambient noise in narrow, constant frequency bands is being exploited.

Hmmm.  Very, very odd comment Arny.  How much a rock weighs has nothing to do with how much weight you can carry.  The signal to noise ratio of a music track is a measurement.  It has nothing to do with what part of it we can hear.

It is another form of FUD so I hope we don't keep using the tactic when we are having an educated and deep technical discussion.

Quote

The noise spectral density is said to have been measured using a 1 Hz bandwidth, but in fact the ear hears in critical bands which are about 1000 times wider around 3.5 KHz. The well  known fallacy of measuring the perceptual qualities of music or ambient noise in narrow, constant frequency bands is being exploited.

Hmmm.  Very, very odd comment Arny.  How much a rock weighs has nothing to do with how much weight you can carry.  The signal to noise ratio of a music track is a measurement.  It has nothing to do with what part of it we can hear.  Indeed  you said it properly yourself: "The musical selection used in the Meridian tests was a good example - the noise floor was only about 70 dB below peak levels and thus easily handled with 16 bits and best practices."

Of course you didn't read the graph right so let's look at a marked up version to make it clear:

 
The embarrassing part here is that I already explained spectral density ~20 pages back. How resistant can a human be to learning anything?

If you thought logically, for just one minute, you'd see that by your wrong method of graphically determining "SNR" you would have to say that 16-bit RPDF results in an "SNR" of over 130 dB given that track and a "noise floor" of about -145 dB.

Anyone can spot this as patently absurd, amirm. You are also misusing the term peak SNR, like all the other "deep technical" () terms which you apparently have no clue what they actually mean.

... I can't even be bothered to decipher the rest of your nonsense.

That's right Amir, in your radical subjectivist world it matters not in the least how well measurements correlate with perception or audibility.

Amir THD and jitter with any number of leading zeroes after the decimal point is what you seem to want to sell and the records of AVS have recorded this any number of times.  How many times has this played out in your posts?

:
THD figure is an improper metric for audibility of distortion. It does not at all follow the rules of psychoacoustics. If I have a 5 Khz signal with a second harmonic at 10 Khz vs a 1 Khz signal with second harmonic at 2 Khz at the same level, the THD will be the same (with respect to the second harmonic). Audibility of these two situations however is very different due to frequency masking and resolution of auditory filters. One must know the spectrum of the harmonic distortions to determine the audibility which of course is never given (or often not measured individually).

We use THD because it used to be easy to measure and has become so ingrained in the industry that it continues to be used. But other than a gross measure, it has little value.

THD is a reasonable metric for electronics and other situations where it can easily be reduced to the point where it is orders of magnitude below inaudible, psychoacoustics notwithstanding.

One leading critic of THD as a metric are Geddes and Lee who are good friends and whose papers I have actually read and recall a few relevant details from. Their criticism of THD is in the context of loudspeakers and other components that still may have audible nonlinear distortion.

Therefore the claim above that "...THD figure is an improper metric for audibility of distortion. " is inconsistent with science, reason and accepted recent research in the field of audio.

That's not the intent for the breed of objectivists of the kind I hang with.

If numbers are irrelevant to audibility, why these  papers?

http://www.gedlee.com/distortion_perception.htm

Hi Amir,

In another thread here, you linked a page with this article on your sales store website, titled "Audibility of Small Distortions" By Amir Majidimehr, the self assessed objectivist/non-hobbyist.

Quote

The “Q” indicates how steep the resonance is in the frequency domain.  In the time domain (not shown), the higher the Q, the more “ringing” the system has.  Ringing means that a transient signal (think of a spike) will create ripples that go on for some time after they disappear.  An ideal system would reproduce that transient with zero ringing.  The higher the Q of a resonance, the more ringing the system has. Reading what I just wrote, if I asked which one of the resonances on the right is more audible, you will likely say High Q.  It seems natural that it has the highest amplitude change and more time domain impact per my explanation.  Yet listening tests show the opposite to be true!  The Low Q is more audible.

Quote from: amirm on 2014-11-27 22:20:59

Here is the measurements from Stuart's article:

What do you make of this and what would the implications be for audibility with and without typical and atypical filters?

Is it possible for out of band resonances to create harmonics in band?

Wow, more quoting posts from other forums in an attempt at making others look as stupid as yourself. That's so pathetic.

No samples arrive at the ear, as you said. The ear receives a continuous mechanical wave that oscillates. Of course the ear doesn't wait for specific filters to arrive, it would have to wait a few seconds if we use a long filter.

As you've quoted many times now, the problem with pre-ringing jj mentions primarily are filter banks in codecs that operate across the whole audible range. He even specifically gives the example of cutoff frequencies of 2 and 4 kHz.

Good morning Arny .

Quote from: Arnold B. Krueger on 2014-11-28 02:46:30

Quote from: Amir link=msg=0 date=

At 3.5 Khz and eyeballing it, the noise floor of the content is around -25 db spl and the peak 90 db.  So the total range is 90+25=115.  How did you get the 70 number?

Nice job of cherry-picking from misleading data.

Boy that is a grumpy way of starting a reply . 

Quote

The noise spectral density is said to have been measured using a 1 Hz bandwidth, but in fact the ear hears in critical bands which are about 1000 times wider around 3.5 KHz. The well  known fallacy of measuring the perceptual qualities of music or ambient noise in narrow, constant frequency bands is being exploited.

Hmmm.  Very, very odd comment Arny.  How much a rock weighs has nothing to do with how much weight you can carry.

The signal to noise ratio of a music track is a measurement.  It has nothing to do with what part of it we can hear.

Indeed  you said it properly yourself: "The musical selection used in the Meridian tests was a good example - the noise floor was only about 70 dB below peak levels and thus easily handled with 16 bits and best practices."

This is what he says about the impact of time domain response of those filters:

"If the filter has substantial energy that leads the main peak, this may be able to affect the auditory system."

So the fact that they are ultrasonic in that regard is not material because of the time domain effect.  I am afraid that torpedoes your arguments throughout this thread.

Wasn't Earl's paper about speakers?

You don't say! 

Quote

As you've quoted many times now, the problem with pre-ringing jj mentions primarily are filter banks in codecs that operate across the whole audible range. He even specifically gives the example of cutoff frequencies of 2 and 4 kHz.

2 to 4 Khz?  Filter Banks?  Sorry no.  Not even close.  The slides could not be clearer.  Let's review them again:

"If the filter has substantial energy that leads the main peak, this may be able to affect the auditory system."

So the fact that they are ultrasonic in that regard is not material because of the time domain effect.  I am afraid that torpedoes your arguments throughout this thread.

Quote

Even a child should understand that ringing (which can be reduced or even eliminated, see previous posts) outside the hearing range is different from clearly within the hearing range.

Hopefully you see that you are completely mistaken here and you have to explain why it is hard to get, not me .

Let's quote Earl, shall we?

This is precisely where the signal-based distortion
metrics fail. In our next paper we will show that .01%
THD of one type of nonlinear system can be
perceived as unacceptable while 10% THD in another
example is perceived as inaudible. Even one of these
simple examples is sufficient to invalidate THD as a
viable metric for discussion of the perception of
distortion. Furthermore, 1% THD is not at all the
same as 1% IM, but we will show that neither
correlates with subjective perception. While some of
the signal-based metrics may be “better” than others,
it is our opinion they all fall short of what we are
seeking.

Any symptoms of dry mouth and restless sleep go with that?

The audibility of resonances is based on research from Dr. Toole and Dr. Olive.  It is based on listening tests of simulated resonances. It was made at low frequencies.

For one thing, the one you have circled on the right is in the ultrasonic/inaudible region.

How the heck would you run a test..

I think that the above almost unbelievable claim is actually something that you believe, Amir. It shows zero insight into the fact that the purpose of audio measurements is to provide some kind of representation of the subjective experience of listening to music through  the equipment with a set of relevant objective measurements.

One of the key concepts of  that I've never been able to get over to you Amir is the concept of comparing comparable items. IOW, compare apples to apples.

Music and background noise are inherently non-comparable.  Music is deterministic and composed of a collection of pure tones. Random noise is not. The energy in random noise is distributed over an infinite number of different frequencies. Accurate measurements of music only require that you measure over the range of frequencies that the music occupies which is often fairly limited. Background noise measurements require measuring over far wider ranges of frequencies. Usually noise measruements are made over the full audio band or over octaves or larger fractions of an octave. Noise measurements over 1 Hz bands are generally useless.

Quote from: amirm on 2014-11-28 18:02:17

Any symptoms of dry mouth and restless sleep go with that?

Nope. No Subjectivist audiophile BS claims about anything on my website or forum escapades, about amp$, wire$, Hi-Re$, etc.
If I like how something looks, feels, etc subjectively. I make no attempt to justify it objectively, be it a Ferrari or an ML amp. I sleep quite well because of it.

Quote from: amirm on 2014-11-27 22:20:59

The audibility of resonances is based on research from Dr. Toole and Dr. Olive.  It is based on listening tests of simulated resonances. It was made at low frequencies.

Well, you're either misinformed, lying, both, or...?
Here is my graph in my article you asked about:



And here is the reference in Dr. Toole's Book:



This is the text for it:

Figure 19.9 shows examples of deviations from fl at for high- (50), medium-
(10), and low- (1) Q resonances at three frequencies when they were adjusted
to the audible threshold levels using pink noise in an anechoic chamber and
for the 200 Hz resonances detected when listening to typical close-miked,
low-reverberation pop and jazz.

So the only data for music is at 200 Hz which is what I referenced.  The rest is for pink noise but even that was tested at max of 5 Khz, not the frequencies you asked me about.

The graph you are showing is a threshold shift between anechoic and reverberant space.  That is a differential score for a different type of test.

Quote

That 12k (linear distortion) hump is certainly within the audible bandwidth, even to someone with as limited hearing as yours.

I don't know Ammar.  Maybe I need the snake oil flat ribbon speaker cable you use with your speakers to hear them.

That aside, the hump in there playing the exact same thing in all the tests at 12 Khz.  In other words it is invariant to the test.  We are not performing a speaker review here Ammar.

Quote from: amirm on 2014-11-27 22:20:59

For one thing, the one you have circled on the right is in the ultrasonic/inaudible region.

Right. That's where one would expect to find Beryllium tweeter breakup resonances.

DG's tests seemed to indicate it might not be with his 2 tweeter sample. But it's not up to me or my side to do anything. It's up to your Hi-Re$ peddler camp to show that the system used in the BS test is transparent to the test and not generating false positives...and there is zero data regarding this.

If you get the measurement wrong as you did, then the interpretation becomes completely false.  Which is what happened when you used the 70 db number to proceed to tell us there is 12 to 13 bits of data in there.  If you start with the right measurement value of 115 db, then you get the right bit depth which is 19.

And that is what Stuart has done.  You see how there is only one set of axis there?  Everything is normalized to the same two axis.

"This [playback] level was chosen for comfort, and because it was high enough for details to be audible but also low enough that 16-bit RPDF dither would be inaudible at the listening position [30]."

RPDF dither is noise.  And for it to be plotted correctly to show inaudibility requires that it be converted to the threshold of hearing base on tones. By making sure 1 Hz bins are used for the FFT of the music, we know there is no process gain and hence the numbers can be compared.

The *peak* level there is about 0 dbfs (relative to 105 db fs).  The floor is about -30 db fs.  You are saying that a Beryllium tweeter breaks up playing such low levels?

And show we have.  The Stuart paper has won an award for best peer reviewed paper at this year's conference.  No one is waiting for lay people with no industry experience to approve anything.  No medical research waits for approval of patients.  Somehow you are confused thinking audio field is an exception because you have two ears.  You need to fill the space between them with audio knowledge which doesn't happen by reading forum posts.