Analogue playback Wander

[jkeny]

I see SoundandMotion has shown exactly how much misinformation Amir spreads in his posts

The imaginary component of the complex numbers show the phase which we don't care about. What is shown is just the "real" component which represents the amplitude of the basis function. We don't need to maintain the phase because we don't need to invert back to time domain.

Hmmm... While admitting I have not followed too closely lately (work and Life issues), I rarely use just real or imaginary data alone unless looking for a reactive component. The magnitude reported in an FFT is normally the RSS of real and imaginary data. The magnitude of a complex number is not accurately given by just the real (or imaginary) component; the magnitude is affected by each. I must have missed something crucial...

No, I don't think you missed anything.
Magnitude = sqrt(re(FFT)^2 + im(FFT)^2) = sqrt(FFT * conj(FFT))
Phase = arctan(im(FFT)/re(FFT))

As can be quite clearly seen the imaginary numbers are involved in both FFT magnitude AND phase
Your statement is totally false yet again, Amir!!

 

[amirm]

Let's see what Zwicker & Fastl have to say about roughness (Psychoacoustics - Facts * Models) - your bible AFA psychoacoustics is concerned:

It is everyone's bible John as far as audio science is concerned. Your attitude toward it as noted above shows again you are not a believer in audio science. It is a difficult text to read and understand though. Here is an example:

And later gets more specific about FM modulation: Frequency modulation can produce much larger roughness than amplitude modulation. A strong frequency modulation over almost the whole frequency range of hearing produces a roughness close to 6 asper. Only amplitude modulation of broad-band noises is able to produce such a large roughness.

Only a turntable has "strong frequency modulation" because of its massive jitter as compared to digital. Indeed, digital is so good/has so little FM modulation that anything but the first sidebands are ignored as having no significant amplitude. As a result, it is treated as AM modulation which too has just those one pair of sidebands. From Bruno's AES paper that Mike post (which itself is copying it from Dunn's famous AES paper on jitter):

See that approximation symbol? The justification is what I explained above.

So what you quoted again damns your argument. It says that at high amplitudes of jitter like we have in turntables, the audible effect is much higher which should make sense to anyone who has heard turntable speed variations. And the mathematics show that what we have in digital is similar to FM modulation (due to bessel coefficients shrinking to almost zero for further components of jitter).

I say again, stick to subjectivism John. You can't just dabble in audio science by googling and quoting a line or two from Google docs.

 

[amirm]

Now let's deal with this one:

Let's see what Zwicker & Fastl have to say about roughness (Psychoacoustics - Facts * Models) - your bible AFA psychoacoustics is concerned:

And: "Our hearing system is most sensitive for sinusoidal frequency modulations at frequencies of modulation in the neighbourhood of 4 Hz."

That is very true. The reason is based on temporal masking (time domain masking). In plain english, what ups and downs in amplitude in time domain are we most sensitive to. Note the bolded part: amplitude. Let's show the JNDs of where the 4Hz came from Zwicker and Fastl book:

So we do indeed see the 4 Hz peak. Which makes sense from evolution point of view and invention of languages like English. For best understanding of speech we want the variations to be 1/4 = 0.25 Hz. And that is pretty close to frequency of consonant and vowel combos in English language. And a more useful use: the ideal reverberation time in a room (RT60) which ranges from 0.2 to 0.5 seconds. But we digress.

Back to our graph, look at the footnote. The amplitude swings are 40 db from high to low. How does that happen? Well, if you severely FM modulate your main tone, that can happen as can AM modulation. Once again, we don't have such a case in digital in even the cheapest implementation. Here is a measurement of Schiit Modi 2 that I just did:

This is pretty bad output for a DAC but even this doesn't remotely fit into the narrative you quoted. There peaks of those jitter components is better than -90 db lower than our main tone. Our main tone is not being modulated in level because jitter is that small. And without its modulation, you don't get to run with the line you quoted.

Here is Zwicker on that:

Note what i highlighted. To get any perception of such modulation, you need to have at least 30 db of it. Our music tones in digital never remotely change levels that way.

 

[jkeny]

It is everyone's bible John as far as audio science is concerned. Your attitude toward it as noted above shows again you are not a believer in audio science. It is a difficult text to read and understand though. Here is an example:



Only a turntable has "strong frequency modulation" because of its massive jitter as compared to digital. Indeed, digital is so good/has so little FM modulation that anything but the first sidebands are ignored as having no significant amplitude. As a result, it is treated as AM modulation which too has just those one pair of sidebands. From Bruno's AES paper that Mike post (which itself is copying it from Dunn's famous AES paper on jitter):

See that approximation symbol? The justification is what I explained above.

So what you quoted again damns your argument. It says that at high amplitudes of jitter like we have in turntables, the audible effect is much higher which should make sense to anyone who has heard turntable speed variations. And the mathematics show that what we have in digital is similar to FM modulation (due to bessel coefficients shrinking to almost zero for further components of jitter).

I say again, stick to subjectivism John. You can't just dabble in audio science by googling and quoting a line or two from Google docs.

Eh, you should stick to schlepping as you have no idea about basic maths as seen in the above post where your devastatingly basic lack of knowledge is corrected by from Donh56 & SAM. So stop the pretense that you know what you are talking about

Again you don't comprehend what you are talking about - the quote shows that FM modulation is 6 times (6 aspers) more audibly sensitive than amplitude modulation - it doesn't matter whether it's strong or weak FM modulations, it will be 6 times more audible. Your focusing on the word strong is a typical attempt at deflection from the message contained in the Zwicker quote


Now back to clock close-in PN - as seen on the high resolution FFT I already posted the energy of the close in frequencies is approaching the signal level energy - the lower the offset from the signal, the higher the energy of the these modulations

 

[jkeny]

Now let's deal with this one:

That is very true. The reason is based on temporal masking (time domain masking). In plain english, what ups and downs in amplitude in time domain are we most sensitive to. Note the bolded part: amplitude.

And FM modulation is 6 times more audibly sensitive - so your example is moot

Let's show the JNDs of where the 4Hz came from Zwicker and Fastl book:


So we do indeed see the 4 Hz peak. Which makes sense from evolution point of view and invention of languages like English. For best understanding of speech we want the variations to be 1/4 = 0.25 Hz. And that is pretty close to frequency of consonant and vowel combos in English language. And a more useful use: the ideal reverberation time in a room (RT60) which ranges from 0.2 to 0.5 seconds. But we digress.

Back to our graph, look at the footnote. The amplitude swings are 40 db from high to low. How does that happen? Well, if you severely FM modulate your main tone, that can happen as can AM modulation. Once again, we don't have such a case in digital in even the cheapest implementation.

So you are mixing up FM & AM modulation & muddying everything with further obfuscation dressed as audio science

Here is a measurement of Schiit Modi 2 that I just did:

Don't need to see another flawed measurement & advertisement for ASR

This is pretty bad output for a DAC but even this doesn't remotely fit into the narrative you quoted. There peaks of those jitter components is better than -90 db lower than our main tone. Our main tone is not being modulated in level because jitter is that small. And without its modulation, you don't get to run with the line you quoted.

Here is Zwicker on that:

Note what i highlighted. To get any perception of such modulation, you need to have at least 30 db of it. Our music tones in digital never remotely change levels that way.

Again cherry picking false examples simply to try to obfuscate

We also have the very first claim you made about close in phase noise - it will just produce noise in the digital audio signal
So add this as another factor into what is being heard - each signal & it's harmonics having a hillock of noise around the signal - in a dynamic music signal we have a dynamically changing small hillocks of modulating noise happening concurrently with FM modulation of the signal.

I & others including Bruno Putzeys who you just quoted have heard the change in sound when clocks of low close-in PN are used (yes real close-in PN, not your attempt at passing off the clocks in the Audiophileo as low PN at close-in offsets)

So you can continue with your demonstration of what you euphemistically deem to be audio science, you can continue with your demonstrations of your mathematical ignorance, you can continue with your denial & deflection & obfuscation but you can't change the fact that it's audible.

 

[jkeny]

Anybody who is that flawed in their understanding of basic maths as you are, Amir, should not be preaching about audio science - you have already demonstrated that you are incapable of understanding the concepts let alone the treatment

I see you have not replied to Donh56 & SAM when they corrected this humongous lack in understanding of yours -yet you post on here

Industrial level denial yet again, Amir, yet again!!

Back you go to ASR & explain yourself - people are waiting to hear!

 

[Steve williams]

I see you have not replied to Donh56 & SAM when they corrected this humongous lack in understanding of yours -yet you post on here

seems he was caught with his pants down at ASR by what I read and yes we have yet to see any response or rebuttal. Surely only once Amir might have the humility to admit he was wrong

 

[jkeny]

seems he was caught with his pants down at ASR by what I read and yes we have yet to see any response or rebuttal. Surely only once Amir might have the humility to admit he was wrong

It's not just that he was caught with his pants down - it's that he exposed himself in such an embarrassing way - basic misunderstanding of sixth grade maths

But the real hypocrisy is him trying to pass himself off as some sort of expert in audio science when he is demonstrating in this lack of knowledge that he is incapable of handling audio science.

We know he has displayed such ignorance in the past with his oversampling in FFTs which Opus & I challenged
And further with his packet retransmission in isochronous USB

Neither of which he fessed up to

We know his measurements & conclusions have been shown many times in the past to be erroneous.

If you wanted an example of how NOT to do audio science he is an exemplar!!

 

[NorthStar]

 

[amirm]

Anybody who is that flawed in their understanding of basic maths as you are, Amir, should not be preaching about audio science - you have already demonstrated that you are incapable of understanding the concepts let alone the treatment

I see you have not replied to Donh56 & SAM when they corrected this humongous lack in understanding of yours -yet you post on here

Industrial level denial yet again, Amir, yet again!!

Back you go to ASR & explain yourself - people are waiting to hear!

Had to take the dogs to the vet. I was definitely wrong there and in a very primitive way!

 

[amirm]

And FM modulation is 6 times more audibly sensitive - so your example is moot

Six times zero is zero. As I explained, FM modulation in the case of digital jitter degenerates into AM Modulation. FM modulation would have sidebands going to infinity. We don't get that. We only get a pair and that is it. The reason as I explained is that the level of FM modulation is so tiny that there simply is no more energy past the first sidebands. Here is one of many example measurements:

As we see, there are only one pair: just like what AM modulation would predict. And FM modulation as vanishingly small modulation index.

That aside, you are arguing against your own case anyway. FM modulation is happening in both analog and digital. If FM modulation is worse, and it is happening at such high modulation index in a turntable, you are in a deeper hole explaining why it is not audible to people there. How no one using a turntable is not complaining about "roughness."

Bottom line is that psychoacoustics works. It says close-in noise gets masked heavily. It is for that reason that we tolerate boatload of it in turntables and tape players. It is just the way the ear works John.

You can't undo that with faulty, biased sighted testing.

 

[amirm]

Here is a cool animation that shows what happens as you increase the level of jitter/FM Modulation. Note that at very small amounts at the start of the animation, we still have our center tone. But as we increase the modulation, it all starts to broaden:

Digital systems are more pure even at the start of this animation. What happens later is what we see and eventually hear in analog formats like turntables.

 

[amirm]

Repeat after me: "you are getting sleepy... you are getting sleepy...."

 

[jkeny]

Had to take the dogs to the vet.

Hope they're OK?

I was definitely wrong there and in a very primitive way!

Well that's refreshingly unexpected- thank you!

It's good to get things off your chest.

Now that the confessional box has opened:
Now would it be too much to ask you to correct your misunderstanding of USB packet re-transmission in isochronous USB & that John Swenson did not say this to you?

I suppose correcting your claim that oversampling is used in FFT is asking too much?

 

[jkeny]

Six times zero is zero. As I explained, FM modulation in the case of digital jitter degenerates into AM Modulation. FM modulation would have sidebands going to infinity. We don't get that. We only get a pair and that is it. The reason as I explained is that the level of FM modulation is so tiny that there simply is no more energy past the first sidebands. Here is one of many example measurements:

As we see, there are only one pair: just like what AM modulation would predict. And FM modulation as vanishingly small modulation index.

But we are talking about close in FM modulation, not the sidebands you show at 6.5KHz & 13KHz or whatever it is.

That aside, you are arguing against your own case anyway. FM modulation is happening in both analog and digital. If FM modulation is worse, and it is happening at such high modulation index in a turntable, you are in a deeper hole explaining why it is not audible to people there. How no one using a turntable is not complaining about "roughness."

When a whole spectrum is modulated in the same direction, as happens in analogue wow, it's just a pitch shift.
As I already explained, this is not what happens with digital audio - the clock timings don't shift as a group, as DOn Hills tried to maintain - the timings don't sequentially group into a positive timing shift & then fall back from there into a swing towards a negative timing shift. Do they? Are you saying you agree with Don Hills on this?

Bottom line is that psychoacoustics works. It says close-in noise gets masked heavily. It is for that reason that we tolerate boatload of it in turntables and tape players. It is just the way the ear works John.

Sorry but your analysis is wrong as I have shown & citing psychoacoustics won't help when your premise is wrong

 

[jkeny]

It is everyone's bible John as far as audio science is concerned. Your attitude toward it as noted above shows again you are not a believer in audio science. It is a difficult text to read and understand though. Here is an example:

Hyperbole, Amir in reference to Zwicker & Fastl - it's not the bible as regards audio science & this blindness is a lot of your problem - you don't see the huge role that Auditory Scene Analysis has to play in auditory perception & neither does Zwicker or Fastl.

No mention in their book of: ASA; of Temporal fine structure; of co-modulated masking release; of binaural envelopes & much, much more; in fact Zwicker & Fastl's book is completely blind to ASA

Auditory Scene Analysis theories & techniques is the main methodology through which the audibility of this close in clock PN will be studied - clarity solidity of soundstage & realism of the illusion playback are the audible characteristics that are a natural focus of ASA

I created a thread on ASR about ASA when I was a member there. I posted links to many papers & comments on the papers. One of the first links I posted on that thread was to the website Auditory Neuroscience with a working example of co-modulated masking release (CMR). You are always talking about masking as a reason for how low level signals are inaudible & yet CMR enhances the audibility of low level signals - exactly the opposite of your assertions. Did you ever try this interactive demo?

As I said before, Amir, your approach is 1st order thinking - dealing with the simplest of signals, fixed tones, pattern-free signals - insufficient for dealing with complex signals of music & the auditory processing that is performed on this signal stream.

 

[jkeny]

And once you understand that ASA research is focused on the auditory streaming, it becomes clear why better clarity & more solid soundstage is reported with audio clocks that have LOW close-in phase noise.

There are two things happening when clocks don't have low close-in jitter:
- spectral impurity i.e each tone is just slightly off
- temporal indistinctness i.e each sound object is a slight bit less distinct in timing

Now both of these issues will not be noticed as problems when listening as there are no distortions to focus on - it certainly won;t be noticed when moving from a moderate level of close-in jitter to a higher level of close-in jitter. The reason being that it's not a linear phenomena - it clicks into place at a certain level of temporal accuracy which low jitter clocks gives & outside of that, it just affects the perception of auditory streams. For those who don't know about ASA research, we group the signals we hear into auditory objects & with dynamic sound (like music) dynamically into auditory streams. This grouping is achieved by the temporal synchrony of signals i.e the notes played on an instrument comprise of a fundamental & harmonics - these move together in unison & the relationship between them is maintain throughout the different notes played. When listening to an orchestra we are able to pick out the fundamental & harmonics belonging to an instrument as they move together & maintain their temporal synchrony. If the spectral quality of the tones change slightly & their temporal synchrony blurs slightly our sense of realism & the 3D illusion of a sound stage is diminished.

We actually notice it much more when the spectral & temporal issues are more correct - it clicks into place more like a realistic illusion.

There seems to be great confusion here about jitter - mostly because people are so caught up in the frequency analysis aspect of FFTs that they ignore the time aspect - some even claim that we are not interested in the phase aspect & try to use their maths ignorance to prove this Some even claim that clock edge mistimings move as a group, just like the pitch shifting in analogue wow (I don't know what type of maths ignorance gives rise to this?)

This is jitter in both the tie domain & frequency domain


In the time domain on the left, we can see the spectral & temporal shifting of tones & harmonics that result from jitter - IMO, this is the underlying reason for why close in jitter (at a certain low level) results in more clarity, more solid 3D sound stage, more realism

Why doesn't this happen with analogue wow - because the whole spectrum is being shifted in pitch at the same time (unlike what happens with jitter) & it is still spectrally pure except shift in frequency, i.e it's not blurred in frequency. As a result the temporal relationship between the signal tones remain intact - they are not blurred as happens with jitter. As I said before it's the difference between a macro shift (analogue wow) & a micro shift (sample jitter)

A good analogy in video might be where analogue wow is represented by the video reel speed fluctautions & jitter is individual objects in the frame having slightly different positions from frame to frame (the movement of an object being independent of other object movements in the frame).We might not notice the video wow if it was small enough but we would likely notice the object jitter as it is pervasive & more intrusive as far as our visual perception is concerned.

 

[jkeny]

Just to complete the picture:

You can instantaneously realize that the Phase Noise is "Noise in terms of phase". Then what is Phase. Again, the mathematical definition of phase just come from high school math as shown below.

How would a signal changes when we changes the phase of the signal. Following plots would give you the answer. The red plot and blue plot has the same frequency and amplitude and the only difference is the phase. In plot (c), you would clearly see the difference. If you see a signal in time domain, the phase difference would result in the delay or advance of a signal.

Now let's think about how a signal would vary with phase in frequency domain. Look at the following three plots. The red plot (a) and blue plot (blue) has the same frequency and amplitude and the only difference is the phase. I already mentioned the time domain difference (left column), now let's look at the frequency domain difference (the center column). If you just focus on the peak point you would not find any obvious difference between the blue signal and red signal. But if you look closely into the region right next to the peak point, you would notice pretty outstanding difference between the blue signal and red signal.

Phase noise is a kind of noise of unexpected (unwanted) phase shift of a signal. In the description above, I compare only two signals but in reality the phase of the noised signal keep changing within a certain range. That's why we call it a noise. (If the phase get shifted to a certain value and remain the same all the time as in the example shown above, we would not call it a noise. it is just a phase shift which can easily corrected. I used a case with only two phase just for easy explanation).

And interestingly, from Amir's thread on ASR "High Resolution Audio: Does It Matter?" in which Bob Stuart's paper "the audibility of typical digital audio filters in a high-fidelity playback system.2 is referenced by Amir as follows:

Notice how the pre-ringing is essentially gone in about 0.7 milliseconds (0.9 msec – 0.2 msec) whereas in the sharp filter case, it went on for more than 3 milliseconds. The need for such sharp filters exists because CD's 44.1 KHz and Video's 48 KHz don't allow a lot of room for gentle roll off. Whereas with high sampling rate we can have very gentle filters that avoid this issue as a practical matter.

Again, this is a theory and not necessarily verified in this test but is one of the best explanations we have as to why there was an audible effect.

What did the subjects really hear? Here is how they described the differences in filtering:

It was reported that filtering gave “softer edges" to the instruments, and “softer leading edges" to musical features with abrupt onsets or changes. Echoes, when audible, were identified as being affected the most clearly by the filtering. It was felt that some of the louder passages of the recording were less aggressive after filtering, and that the inner voices (second violin and viola) had “a nasal quality." Overall, the filtered recording gave a “smaller and flatter auditory image,” and specifically the physical space around the quartet seemed smaller.​

The effect of conversion from 24 bits to 16 bits (“quantization”):

Listeners described that quantization gave a “roughness" or “edginess" to the tone of the instruments, and that quantization had a significant impact on decay, particularly after homophonic chords, where “decay was sustained louder for longer and then died suddenly." This could be an effect of quantization distortion; it is interesting that this was audible even in a 24-bit system, and is consistent with the hypotheses of Stuart [29] that 16 bits are not sufficient for inaudible quantization.​

So apparently, extremely low level signals resulting from differences in filters @24Bits are audible & described in that first excerpt & the effects of 16 bit quantisation noise is audible & described in the second quote.
Note that none of these descriptions mention "noise" they describe auditory scene analysis characteristics of the sound streams - such as “smaller and flatter auditory image,”; "“softer edges" to the instruments"

 

[jkeny]

Hope they're OK?

Well that's refreshingly unexpected- thank you!

It's good to get things off your chest.

Now that the confessional box has opened:
Now would it be too much to ask you to correct your misunderstanding of USB packet re-transmission in isochronous USB & that John Swenson did not say this to you?

I suppose correcting your claim that oversampling is used in FFT is asking too much?

I guess that was expecting too much from you, Amir?

 

[amirm]

I guess that was expecting too much from you, Amir?

No, it is not that. I just don't have much interest in interacting with you on what amounts to a fist fight all the time.

Anyway, since you won't stop, here are your answers:

Now that the confessional box has opened:
Now would it be too much to ask you to correct your misunderstanding of USB packet re-transmission in isochronous USB & that John Swenson did not say this to you?

First, there is no misunderstanding on my part on how USB works. I have written hundreds of posts and I challenge you to find one that says USB has retransmission. Indeed I have written extensively about Internet streaming where I do mention it there. Here is an article I wrote for Widescreen Review Magazine two to three years ago on USB: http://audiosciencereview.com/forum...performance-pc-server-interfaces-async-usb.8/. Again, there is no mention of retransmission. So your contention that I don't know this topic is flat wrong.

As to what John said about Regen, here it is. I asked him why he thought putting a USB hub/repeater in the path of audio would be helpful. After all, the stronger you make the digital signal, the more chance it has to bleed into the sensitive analog circuits. He said that he had designed a USB core once and that there was an error mitigation mechanism in the receiver which would tell the transmitter that there were errors. And that some implementations (?) would cause the transmitter to strengthen its signal and hence cause more noise problems. I took his word for it as it is not something that I know about.

The punchline was that he said he had realized USB 3.0 controllers that are now very common have done away with this feature altogether so the problem he thought he was solving, was no longer an issue!!!

That is the extent of my conversation with him on this bit. As to the word "retransmission" I had a mental block on the word "mitigation" when I wrote the post that you have been running with all this time. That was my bad and I backed off when I said it. Neither he, nor I, think that there is retransmission in USB.