[theveterans]

@sunjam
For fun, wanna train your ears to tell the difference between Sinc-L with 7th order noise shaper vs Normal poly sinc lp with TPDF dithering using Goldensound files?

BTW, sighted listening this, I can clearly hear the difference between the two. I'm currently training my ears so that when I load these puppies up on foobar ABX, I'll get at least 15/20 or better though I'm hoping to beat goldensound with a 20/20 correct guesses

Right now, not even looking at which filter version is currently playing in Roon as I'm typing this, I know already which one is Sinc-L and which one is the poly sinc lp filter

https://drive.google.com/drive/folders/18EUi9cYeHN4BF9fn5v2ELQPSNdBJkRmw

 

[theveterans]

I must say, the original stream from Qobuz sounds exactly like the normal filter than the high performance filter. Yes the normal filter is the one that's truer to original qobuz 44.1 KHz stream with sighted listening

 

[sunjam]

Thanks for your reply. I shared most of the things you mentioned especially if it was modified as below:

You can degrade anything until it sounds bad. That isn't what standards are for. They aren't for best case scenarios either. Standards are designed to perform well for their intended purpose. The purpose of home audio is to play high fidelity music in the home. Redbook CDs take that one step further. They produce a signal that is indistinguishable from the original. The dynamic range, frequency response, distortion and timing are all good better than human ears can hear. That means you don't have to reinvent the wheel or build in a little buffer just in case. All you need to do is play the digital file through a modern inexpensive player or DAC that is performing to spec and you're golden good.

You can dive into theory to try to invent find out reasons why it MIGHT not be good enough, and you can think up ways to degrade upgrade the signal to insure it doesn't sound good better. But all that is mental monkey spanking, because out of the box 16/44.1 is perfect to human ears. It's all you need.

I don't know why you guys bother reading sunbeam's sunjam's blather comments. There's nothing something to be gained from it. I wouldn't reply to goosenaught except that he slipped and actually started making some valid comments. I don't hold grudges, I just give the respect a person deserves, or doesn't as the case may be. I'm sure I'll be rewarded with disrespect soon from our tag team professional wrestlers.

 

[sunjam]

Oh why did we have CD players in the 1980s if 16bit (or 14bit) wasn't adequate enough?

 

[castleofargh]

i will go with this figure since i actually dont remember if pipewire uses noise shaping
but i actually didnt know that noiseshaping pushes the noisefloor below -96db with cd quality, i guess the requirement is to still process in 24 bit with a 16bit file?

i know masking, and you are simply not "sensitive" enough while listen to 80db to perceive tones at -80db, but these -8db tested seem to give us a "absolute" figure of what "could" influence what we hear and with a normal listening level of 80db we land at -88db, this should atleast tell us that cd quality is pretty much "spot on" with theoretical noise level

personally i dont think noise levels do much, distortion is far more important to keep under control,
but i still question whether a "smooth sounding noise" might influence things more than a "irritating sounding noise" in terms of how we process the noise level subconsciously, even if its only barely perceptible
i definitely noticed that with my new BLA interface 1. the noise when you hear it actually sounds way smoother/pleasent than others 2. when i listen with -20-30 db on the interface i actually think the background becomes less "dark" ... i still dont hear "the noisefloor on its own" but it seems to influence how i perceive the music overall, its sounds a bit more "veiled"

if i go by the around -105db noisefloor or so of the BLA interface this means i landed at around -60-80db, before "it seems" like the higher noisefloor influences the music, tho keep in mind this is just my own theory here...

Also regarding headroom: i use a 10 band eq, each band is only subtracting frequencys, instead of a lowshelf filter to boost lows i use a highshelf filter to reduce highs for example... i use some higher q peak filters but these should nowhere go above +1-2db
now to get a around -3db peak level (with max pipewire volume) on the interface (it shows with lights) i actually have to set a pregain before the eq at -12db, i noticed the same with my aune x8 before, it simply sounds distorted if i dont set the pregain that low

i guess the imortant part is " the bla interface gets -3db" and "the noisefloor is at around -105db" unless i look into non-noise shaped dither, then my -12db pregain + normal dither with cd quality lands me at a around -80-82db noisefloor on max volume, if we account for my housecurve which further reduces highs these will be at -76-80db or so

now with these -76-82db noisefloor figures i get theoretically on my setup, these still seem very much fine, but if i would reduce the interface by further -10-20db and max my amp volume so i land around the same volume i will start to perceive a "brighter" background, i guess around -60 db actually

i think a real life example shows best why cd quality "might not be enough" in terms of headroom tho im still fine with my current setup, i just still doesnt seem to get why i need so much pregain reduciton ( i dont think something is messing around with volume levels, the plugin/program i use for overall volume also doesnt let you go above 100% volume, which some older pulseaudio plugins let you do )

EDIT: even with disabling EQ, i need a substantial reduction in volume to get no clipping lights on the interface, and i noticed the same thing audibly with my aune X8 before which doesnt show levels (i used deezer/firefox with max volume here, linux volume at 100% too and the +12db light on the interface is blinking up)

1/ sighted impressions that look very much like explanations one could have without doing any test at all.
2/ I don't understand what that BLA interface dBu value is supposed to say. You loop the output of a DAC into its input? It is the DAC and if so are you sure the light tell you about clipping instead of just voltage value?

 

[gregorio]

To make it clear again, I am at most at "primary level" for audio science. Having said that, I thought we discussing facts related to audio science here. Am I correct?

No you are not correct, just for a change. You are not discussing “facts related to audio science”, the rest of us are but you are doing the opposite and promoting pseudoscience related to audio science and falsely calling it “facts”.

I don't think people who have critical thinking would consider a primary school kid cannot show any fact in his comments for an on-line discussion. Have you ever encountered any kid who is considered as "smart kid"?

No, I’ve never met or even heard of a primary school kid who knew more about digital audio than a MIT graduate with a masters degree degree in computer science and a bachelor’s degree in electrical engineering who invented certain widely used audio codecs (Monty). Please give us an example of one.

Given the definition of Brownian motion as shown about, do you think 0.63 micro pascal is possible? Does 0.63 micro pascal really not exist? Hmmm...
I can see at least two situations that could exist
1. close to vacuum
2. cold enough to slow down the random motion of particles

How could those two situations exist? If the listener has died from hypoxia or has frozen to death, how do they still exist? Having said this, maybe situation #1 can exist, as your responses are indeed symptomatic of hypoxia and what is between your ears certainly appears to be “close to a vacuum”! lol

The more I digged into the influence of the misleading claims in the Monty's video, the more I found it is terrifiying .....

I’m sure. The actual demonstrated facts would indeed appear “terrifying” to someone trying to peddle pseudoscience, and they would need to lie and falsely call it “misleading”!

Concept 1: there is an absolute minimum sound (e.g. the "min" dB)
Concept 2: you cannot hear the difference between Y dB and Y + 'min' dB because the absolute minimum sound is 'min' dB
Is concept 2 supported by concept 1?

If Y dB is below min dB and you cannot produce/reproduce Y dB, then there is no “Y + min dB”, there’s just min dB! And how can you claim to hear a difference between min dB and the same min dB?

So to summarise your argument: “Your knowledge, which is equivalent to a smart kindergarten child, dictates that the difference between hi-res and CD exists and can be experienced by an audiophile providing they’re dead.” - That definitely seems to me to be the result of “critical thinking”! ROFL

G

 

[castleofargh]

Having sunjam claim to use critical thinking against pseudo-science is like those 'freedom something' groups that turn out to seek anything but freedom for their people.
Also, I feel his genuine desire to be heard, but he has nothing to say. It's weird, and his dishonesty to try and invent content is a PITA.

 

[sunjam]

(The following is my latest writing regarding the Hi-Res vs CD format. As you guys are knowledgeable in audio science, I would like to see if you have any comment about it. Thanks)​

What's the difference between CD-quality music and Hi-Res music.​

Some people on various on-line audio science forums claim that "Hi-Res is useless" and "you cannot hear the difference between the CD-quality music and Hi-Res music".

You may ask, "Really? Is Hi-Res music really useless?"

Well. Let's check together with our critical thinking!

(note: this is a high level comparison between CD-Quality format and Hi-Res format. It should give you a good starting point for your own investigation)

Reconstruct music from CD-quality format

The following graph (Figure 1) shows the original analog input signal (a perfect sine wave)


Figure 1: Original analog input signal (a perfect sine wave


As the final audio signal reconstruction process can never be perfect, the reconstructed analog signal looks like a "smooth" sine wave but, in fact, it is smeared due to 16-bit quantization and other artifacts introduced during the ADC and DAC process as shown on Figure 2 below:

(note: in Monty's video, it claims that "the analog signal can be reconstructed "losslessly, smoothly, and with the exact timing of the original signal [by using CD fromat]" which I disagreed)


Figure 2: the reconsturcted sine wave is smeared due to 16-bit quantization and other artifacts introduced in the ADC and DAC process (simulated to show the effects of reconstruction artifacts).
Note 1: Sorry for my poor hand-drawing, the graph should be a "smooth" sine wave with smearing.
Note 2: the level of smearing may not be in scale with the actual output from a DAC with "perfect" filter as the above graph is for indicating the effects of artifacts in the reconstruction process)
Note 3: Dithering is one of the source of the smearing.

Does Hi Res help to reconstruct better music?

Let's look at the following graph (Figure 3) that shows the ouput of the same analog input signal. (The graph is a simulated graph with 24-bit quantization. The smearing is reduced due to 24-bit quantization. As you can see here, the degree of smearing is less than the one shown on Figure 2 for CD-quality format).



Figure 3: the reconsturcted sine wave is smeared due to 24-bit quantization and other artifacts introduced in the ADC and DAC process (simulated to show the effects of reconstruction artifacts). The degree of smearing is much less than the one with 16-bit quantization that is used by the CD-quality format.
Note 1: Sorry for my poor hand-drawing, the graph should be a "smooth" sine wave with less smearing than the 16-bit version.
Note 2: the level of smearing may not be in scale with the actual output from a DAC with "perfect" filter as the above graph is for indicating the effects of artifacts in the reconstruction process
Note 3: Dithering is one of the source of the smearing.

Why Hi-Res has less smearing when compared with CD-quality format?

For Hi-Res, there are 16,777,216 different levels (24-bit) to digitize the amplitude of the analog input signal;

For CD format, there are only 65,536 different levels (16-bit) to digitized the amplitude of the same analog input signal (i.e. Hi-Res is 256x better than CD format in terms of number of levels for digitizing the amplitude of the analog input signal).

Which one you need? CD format or Hi-Res?


It all depends on if you can hear the difference between the reconstructed audio signal in Hi-Res format (the blue one with smearing) and the one in CD-quality format (the red one with worser smearing).

If you cannot hear the difference, CD format would probably be good enough for you. However, if you can hear the difference, Hi-Res would help your system to reconstruct better final audio output for your enjoyment.

No matter what format you use, enjoy your good music and let's debunk pseudo science together with our critical thinking. Yeah!

 

[sunjam]

Reasoning fallacy example?:

Let's look at the following:

@gregorio is attempting to show with scientific evidence that there is an absolute minimum sound.

I don't know how this is related to the scientific evidence about "people cannot hear the difference between Hi-Res or CD" <=== I am not sure if @gregorio want to support such claim with the idea of "absolute minimum sound"

Concept 1: there is an absolute minimum sound (e.g. the "min" dB)
Concept 2: you cannot hear the difference between Y dB and Y + 'min' dB because the absolute minimum sound is 'min' dB

Is concept 2 supported by concept 1?

If Y dB is below min dB and you cannot produce/reproduce Y dB, then there is no “Y + min dB”, there’s just min dB! And how can you claim to hear a difference between min dB and the same min dB?

@gregorio What if Y db is not below 'min' dB

"I don't know how this is related to the scientific evidence about "people cannot hear the difference between Hi-Res or CD" <=== I am not sure if @gregorio want to support such claim with the idea of "absolute minimum sound". <== So, in your mind, it is a supporting argument?

 

[71 dB]

Brownian motion and pressure:

"This is the first time I see Brownian motion mentioned in acoustics" <=== same here. I found his idea is pretty interesting.

Let's look at what's Brownian motion:



I remember you were asking why "wouldn't -30 dB SPL and 0.63... micro pascals exist" as shown below:

And @gregorio was explaining to you with the idea of Brownian motion.
Given the definition of Brownian motion as shown about, do you think 0.63 micro pascal is possible? Does 0.63 micro pascal really not exist? Hmmm...

I can see at least two situations that could exist
1. close to vacuum
2. cold enough to slow down the random motion of particles

Sound is pressure deviations from static pressure (typically 1 bar or so) on frequency band of human hearing. Pressure deviation goes to zero multiple times per second as the pressure deviation changes from negative to positive and vice versa. In that sense the momentary sound pressure (pressure deviation) can be arbitrarily small, even zero, but due to Brownian motion it can't STAY below certain levels which in typical room temperatures, air pressure etc. seems to be about -23.7 dBSPL or 1.3 𝜇Pa. Changing the conditions change the effect of Brownian motions, but we hardly want to listen to music at temperature of -235°C (-391°F) which is what would drop the Brownian motion to 0.63 Pa if I calculated correctly. Or how about dropping the air pressure to about half a bar? Is that the secret to breathtaking audio?

BTW, I thought the Ideal Gas Law is taught in high school physics. No?
Or the high school teachers, these days, teach a pseudo science version of the Ideal Gas Law in high school? i.e. the "correct but not absolute" version of the Ideal Gas Law?

I don't know about any pseudo science version of it, sorry. I was in high school in late 80s. We didn't watch Youtube videos obviously. We learned stuff from textbooks. I don't know what they teach nowadays.

 

[sunjam]

I don't know about any pseudo science version of it, sorry. I was in high school in late 80s. We didn't watch Youtube videos obviously. We learned stuff from textbooks. I don't know what they teach nowadays.

Cool, thanks a lot for your reply. I would believe textbooks are much reliable than YouTube video (even from the so-called expert).

Nowadays, students are watching a lot YouTube video in high school during classroom time. I was shocked to know that they use the Monty's video for teaching. I am very worried about our next generation. Why mis-lead them with "valid but not absolute" claims. Sigh...

Note: The one from Archimago is a little bit better. He has a qualifier stated ("The exception being non-oversampling DACs that specifically chooe to go down the stair-step path")

 

[71 dB]

Some people on various on-line audio science forums claim that "Hi-Res is useless" and "you cannot hear the difference between the CD-quality music and Hi-Res music".

The fact that there are plenty of people saying "Hi-res is useless" is telling. If Hi-res was a real improvement in audio quality, why would anyone say it isn't? Sure, people with bad eyesight may say Blu-ray isn't better than DVD, but we are not talking about half-deaf people! People with very good hearing (and often proper education/knowledge) are saying it!

You may ask, "Really? Is Hi-Res music really useless?"

For music consumers it is. For music production it may have some benefits (mostly the bit depth part, but potentially also bigger sample rate).

Reconstruct music from CD-quality format

The following graph (Figure 1) shows the original analog input signal (a perfect sine wave)


Figure 1: Original analog input signal (a perfect sine wave

This is a digital version of the "original" analog signal. It is made of pixels. It just looks pretty "analog" thanks to the resolution being somewhat high. Pictures of signals can be misleading, because we hear sound differently from seeing the pictures of signals. Eyes and ears are quite different senses. Pictures of signals can however be very informative if we interpret them correctly. This usually takes experience and education most people simply don't have.

As the final audio signal reconstruction process can never be perfect, the reconstructed analog signal looks like a "smooth" sine wave but, in fact, it is smeared due to 16-bit quantization and other artifacts introduced during the ADC and DAC process as shown on Figure 2 below:

(note: in Monty's video, it claims that "the analog signal can be reconstructed "losslessly, smoothly, and with the exact timing of the original signal [by using CD fromat]" which I disagreed)


Figure 2: the reconsturcted sine wave is smeared due to 16-bit quantization and other artifacts introduced in the ADC and DAC process (simulated to show the effects of reconstruction artifacts).
Note 1: Sorry for my poor hand-drawing, the graph should be a "smooth" sine wave with smearing.
Note 2: the level of smearing may not be in scale with the actual output from a DAC with "perfect" filter as the above graph is for indicating the effects of artifacts in the reconstruction process)
Note 3: Dithering is one of the source of the smearing.

16-bit quantization (with dither) alone DOES NOT smear the signal at all as long as the signal is properly bandlimited. What happens is dither gets summed with the signal. ADCs and DACs can cause smearing if they are not ideal which they of course are not, but the theory of digital audio doesn't set any upper limits to how close to ideal ADCs and DACs can be. We only need them to be audibly ideal.

Does Hi Res help to reconstruct better music?

Let's look at the following graph (Figure 3) that shows the ouput of the same analog input signal. (The graph is a simulated graph with 24-bit quantization. The smearing is reduced due to 24-bit quantization. As you can see here, the degree of smearing is less than the one shown on Figure 2 for CD-quality format).



Figure 3: the reconsturcted sine wave is smeared due to 24-bit quantization and other artifacts introduced in the ADC and DAC process (simulated to show the effects of reconstruction artifacts). The degree of smearing is much less than the one with 16-bit quantization that is used by the CD-quality format.
Note 1: Sorry for my poor hand-drawing, the graph should be a "smooth" sine wave with less smearing than the 16-bit version.
Note 2: the level of smearing may not be in scale with the actual output from a DAC with "perfect" filter as the above graph is for indicating the effects of artifacts in the reconstruction process
Note 3: Dithering is one of the source of the smearing.

There is no smearing due to quantization!* What we gain from using 24 bits instead of just 16 is the dither noise added to the signal is 48 dB lower. In technical sense this is an improvement, but in audible sense it is not, because 16 bit dither is already below audibility. This is why 24 bit can be considered beneficial in music production, but not for music consumers.

* Assuming properly bandlimited signal + dither.

Why Hi-Res has less smearing when compared with CD-quality format?

For Hi-Res, there are 16,777,216 different levels (24-bit) to digitize the amplitude of the analog input signal;

For CD format, there are only 65,536 different levels (16-bit) to digitized the amplitude of the same analog input signal (i.e. Hi-Res is 256x better than CD format in terms of number of levels for digitizing the amplitude of the analog input signal).

Which one you need? CD format or Hi-Res?

It all depends on if you can hear the difference between the reconstructed audio signal in Hi-Res format (the blue one with smearing) and the one in CD-quality format (the red one with worser smearing).

If you cannot hear the difference, CD format would probably be good enough for you. However, if you can hear the difference, Hi-Res would help your system to reconstruct better final audio output for your enjoyment.

No matter what format you use, enjoy your good music and let's debunk pseudo science together with our critical thinking. Yeah!

There is no smearing due to the quantization. There is dither noise. 16 bit dither is inaudible. 24 bit is 48 dB more inaudible. It is literally an improvement we can't hear. There is a threshold after which adding bits doesn't improve the sound audibly. In my opinion this threshold is 13 bit. Some people may say 12 bit. Some people may say 14 bit, but it is somewhere there. It is very clear 8 bit is not enough and 10 bit just doesn't do it. 16 bit is clearly on the safe side. There is definitely not need for more than that in consumer audio. Clearly 24 bit is ridiculous overkill in consumer audio.

Sure, 16,777,216 looks so much better than 65,536. Of course it does. 16,777,216 on my bank account would look AWESOME while 65,536 is much less than what I have in savings. However, it takes understanding/knowledge of digital audio and human hearing to know what it really means. 65,536 levels is already enough! In fact 2^13 = 8192 levels is enough! This may seem surprising, but your pictures of "analog" smooth signals are just pictures of a few hundred pixels wide and tall. You didn't need millions of pixels wide pictures, did you? Create a 8192 pixels high picture of signals and you'll understand how "smooth" they are, 4 times taller than 4K monitors! Yes, we see signals differently from how we hear them, but maybe this helps you believe 8192 levels is enough in consumer audio. Now compare 65,536 to 8192. Looks big doesn't it? Suddenly 16 bits looks pretty impressive, doesn't it?

Dither cleverly transforms the quantization error into uncorrelated noise. We have 100 % pure signal + dither noise. All we need is enough bits to have the dither noise be quiet enough to be inaudible. Yes, thats about 13 bits! That's it. Pure undistorted signal with inaudibly quiet noise with just 13 bits. With 16 bits the dither is 18 dB lower! That's why 65,536 is more than enough. Obviously 16,777,216 is ridiculous overkill.

 

[71 dB]

Cool, thanks a lot for your reply. I would believe textbooks are much reliable than YouTube video (even from the so-called expert).

Youtube videos can of course be just as reliable as textbooks depending on who makes them and for what purpose. Students may find Youtube videos more pleasant and engaging way of learning. I have learned a lot of things watching Youtube videos (for example music theory which I didn't learn at all in school).

Nowadays, students are watching a lot YouTube video in high school during classroom time. I was shocked to know that they use the Monty's video for teaching. I am very worried about our next generation. Why mis-lead them with "valid but not absolute" claims. Sigh...

I watched the Monty video some years ago. I don't remember it 100 %, but I don't recall anything wrong with it. I wonder what about it shocks you so much...

Note: The one from Archimago is a little bit better. He has a qualifier stated ("The exception being non-oversampling DACs that specifically chooe to go down the stair-step path")

The signal level of that is something like -72 dBFS. That is incredibly quiet. If you listen to levels where the peaks (assume 0 dBFS) go to 100 dB, this signal is 28 dB which means a quiet living room. The noise from AC is likely to mask this signal even when listening to such a loud levels. Also when listening to loud music, your hearing threshold gets temporarily raised. I ask you to think critically how accurately you need signals this quiet to be reconstructed? Now compare that to how accurately they are constructed.

 

[sunjam]

The fact that there are plenty of people saying "Hi-res is useless" is telling. If Hi-res was a real improvement in audio quality, why would anyone say it isn't? Sure, people with bad eyesight may say Blu-ray isn't better than DVD, but we are not talking about half-deaf people! People with very good hearing (and often proper education/knowledge) are saying it!


For music consumers it is. For music production it may have some benefits (mostly the bit depth part, but potentially also bigger sample rate).


This is a digital version of the "original" analog signal. It is made of pixels. It just looks pretty "analog" thanks to the resolution being somewhat high. Pictures of signals can be misleading, because we hear sound differently from seeing the pictures of signals. Eyes and ears are quite different senses. Pictures of signals can however be very informative if we interpret them correctly. This usually takes experience and education most people simply don't have.


16-bit quantization (with dither) alone DOES NOT smear the signal at all as long as the signal is properly bandlimited. What happens is dither gets summed with the signal. ADCs and DACs can cause smearing if they are not ideal which they of course are not, but the theory of digital audio doesn't set any upper limits to how close to ideal ADCs and DACs can be. We only need them to be audibly ideal.


There is no smearing due to quantization!* What we gain from using 24 bits instead of just 16 is the dither noise added to the signal is 48 dB lower. In technical sense this is an improvement, but in audible sense it is not, because 16 bit dither is already below audibility. This is why 24 bit can be considered beneficial in music production, but not for music consumers.

* Assuming properly bandlimited signal + dither.


There is no smearing due to the quantization. There is dither noise. 16 bit dither is inaudible. 24 bit is 48 dB more inaudible. It is literally an improvement we can't hear. There is a threshold after which adding bits doesn't improve the sound audibly. In my opinion this threshold is 13 bit. Some people may say 12 bit. Some people may say 14 bit, but it is somewhere there. It is very clear 8 bit is not enough and 10 bit just doesn't do it. 16 bit is clearly on the safe side. There is definitely not need for more than that in consumer audio. Clearly 24 bit is ridiculous overkill in consumer audio.

Sure, 16,777,216 looks so much better than 65,536. Of course it does. 16,777,216 on my bank account would look AWESOME while 65,536 is much less than what I have in savings. However, it takes understanding/knowledge of digital audio and human hearing to know what it really means. 65,536 levels is already enough! In fact 2^13 = 8192 levels is enough! This may seem surprising, but your pictures of "analog" smooth signals are just pictures of a few hundred pixels wide and tall. You didn't need millions of pixels wide pictures, did you? Create a 8192 pixels high picture of signals and you'll understand how "smooth" they are, 4 times taller than 4K monitors! Yes, we see signals differently from how we hear them, but maybe this helps you believe 8192 levels is enough in consumer audio. Now compare 65,536 to 8192. Looks big doesn't it? Suddenly 16 bits looks pretty impressive, doesn't it?

Dither cleverly transforms the quantization error into uncorrelated noise. We have 100 % pure signal + dither noise. All we need is enough bits to have the dither noise be quiet enough to be inaudible. Yes, thats about 13 bits! That's it. Pure undistorted signal with inaudibly quiet noise with just 13 bits. With 16 bits the dither is 18 dB lower! That's why 65,536 is more than enough. Obviously 16,777,216 is ridiculous overkill.

Cool, thanks for your comment.

"Dither cleverly transforms the quantization error into uncorrelated noise [in the final audio output]." <== I agreed

With dithering, where the "uncorrelated noise" goes? IMO, it is shown as the smearing of the final audio output. For this, I want to confirm my understanding with you guys.

Dither - Wikipedia

 

[71 dB]

Cool, thanks for your comment.

"Dither cleverly transforms the quantization error into uncorrelated noise [in the final audio output]." <== I agreed

With dithering, where the "uncorrelated noise" goes? IMO, it is shown as the smearing of the final audio output. For this, I want to confirm my understanding with you guys.

Dither lives with the signals, obviously. Visually it does smear the signal, but we don't hear it that way even if it was loud enough to be heard. Take a recording of music for cello and piano (say, a cello sonata). If you look at the signal, clearly sound from piano and cello smear each other badly, but we hear the instrumeats clean and separate. Our brain is able to do that separation. Similarly our brain is able to tell apart the signal and dither if the dither was loud enough to be heard (which it is not at 16 bit). Since dither is inaudible, all we hear is the signal. Visual pictures of smeared signal doesn't matter.