Distortion down to -300 dB, what exactly does that mean physically?

[audio2design]

Good point, the 300+ db refer to Verilog simulations.

Shouldn't such simulations account for thermal noise?

Depend on whether you are using purely digital Verilog or Verilog -A or AMS or ..

However, if you convinced yourself that you can hear signals 300db down because you correlated a listening test to a math result, one can conclude two things. You are delusional coupled with a super ego about your abilities, you are bad at math and have an inability to estimate and work with large numbers, or you screwed up the implementation and accidentally generated a rounding error by using math of insufficient precision for intermediate values.

 

[Andysu]

REW only goes down to -240dB

Wow Spectrum Lab goes down to -1150dB I more less stopped there as it seemed to be almost endless. Designed for radio astronomy, great for LFE.1 sub bass waterfalls and lot more.

 

[JohnA]

There is a reference to -350db from the PGGB guys
https://audiophilestyle.com/forums/...eady-brew-of-math-and-magic/page/23/#comments


"the noise shaper PGGB uses for 16FS signal to noise shape output signal to 32bits has a noise floor below -350dB in the audible range and can easily reproduce a tone at -200dB anywhere in the audible range. "

This is the first time I come across such a reference outside Rob's videos/posts.

Could there be something into this?
I keep an open mind when it comes to human perception.


(Just acting like the devil's advocate, I'm aware that audiophooling is a more likely explanation)

 

[Spkrdctr]

Depend on whether you are using purely digital Verilog or Verilog -A or AMS or ..

However, if you convinced yourself that you can hear signals 300db down, one can conclude two things. You are delusional coupled with a super ego about your abilities.

What I call the "God" complex. Men thinking they have the hearing of a God. Men can't even hear what a common cat or dog hears. Our hearing is LOUSY. A well known scientific fact. People are multi-tools. Fair vision, smell and taste. In hearing we are not even fair, we are lousy. Hearing is our WORST ability. That's the all important bottom line in all of these discussions. All you Gods need to join the rest of the common humans on this planet, especially you audiophiles!

 

[j_j]

This is a wrong equation.
1 bit = 6 dB<- this is a lie

What? If you're upset he didn't say 6.02, ok, I'll give you that.

 

[j_j]

How do we know what he can hear/perceive

It's simple. Air is made of N2, O2, Argon, and small amounts of CO2, H2O, and other stuff. "Air pressure" is created by those molecules individually bouncing off the surfaces of things. INDIVIDUAL MOLECULES doing that.

This means, at the lowest level, air pressure consists of noise. Because the "averaging" is enormous, the DC component is what we call air pressure. Obviously.

But, now, let's talk about the AC component. Without going into a whole lot of mathematics that is impossible to type in ASCII, the answer is

"For an average sized eardrum, the noise due to the molecular nature of air is somewhere around 6 to 8.5 DB SPL white noise." Yeah. Really.

Now detectability of noise goes by how much energy there is in every ERB, an ERB being the cochlear filter bandwidth around any chosen frequency between about 20Hz and 19000Hz (hearing above that is limited to the very entrance to the basilar membrane and the filter bandwith at the entrance). At low frequencies, an ERB is about 40Hz wide, give or take. When 1/4 octave is wider than 40Hz, then it grows more or less by 1/4 octaves.

Using all that established knowledge, we figure out that the atmospheric noise is JUST below the absolute threshold of hearing at the most sensitive point in an undamaged person's ears.

Using other established knowledge, we know that a tone is masked by noise in an ERB when it's about 5.5 dB lower than the noise in the ERB. There are some circumstances that may make this number 8dB.

So, let's make this 10dB (being conservative). Using a very narrow estimate for ERB's (that has to be over conservative), the noise floor in the most sensitive band is at about -20dB SPL. This means that detecting tones below -30dB SPL is, frankly, not going to happen.

So, yes, Peter, we can say that, and I just did.

 

[j_j]

Mea culpa, maximum dynamic range if you're OK with basically instant death is 214 dB SPL. Thanks, my quick search turned up the wrong number.

This reminds me of a discussion that just happened in another forum, where somebody suggested that PCM should use 64 bit fixed-point.

Setting the 0 dB point at 6dB lower than the noise floor of the atmosphere, we get to a peak level of approximately 2 GIGA ATMOSPHERES.

So, no, no, no. Look up the pressure inside the sun for comparison.

 

[Blumlein 88]

It's simple. Air is made of N2, Argon, and small amounts of CO2, H2O, and other stuff. "Air pressure" is created by those molecules individually bouncing off the surfaces of things. INDIVIDUAL MOLECULES doing that.

This means, at the lowest level, air pressure consists of noise. Because the "averaging" is enormous, the DC component is what we call air pressure. Obviously.

But, now, let's talk about the AC component. Without going into a whole lot of mathematics that is impossible to type in ASCII, the answer is

"For an average sized eardrum, the noise due to the molecular nature of air is somewhere around 6 to 8.5 DB SPL white noise." Yeah. Really.

Now detectability of noise goes by how much energy there is in every ERB, an ERB being the cochlear filter bandwidth around any chosen frequency between about 20Hz and 19000Hz (hearing above that is limited to the very entrance to the basilar membrane and the filter bandwith at the entrance). At low frequencies, an ERB is about 40Hz wide, give or take. When 1/4 octave is wider than 40Hz, then it grows more or less by 1/4 octaves.

Using all that established knowledge, we figure out that the atmospheric noise is JUST below the absolute threshold of hearing at the most sensitive point in an undamaged person's ears.

Using other established knowledge, we know that a tone is masked by noise in an ERB when it's about 5.5 dB lower than the noise in the ERB. There are some circumstances that may make this number 8dB.

So, let's make this 10dB (being conservative). Using a very narrow estimate for ERB's (that has to be over conservative), the noise floor in the most sensitive band is at about -20dB SPL. This means that detecting tones below -30dB SPL is, frankly, not going to happen.

So, yes, Peter, we can say that, and I just did.

I've tried to explain this to so many people. And they either don't understand it or choose not to. I don't have the credibility that J_J has so too many people just ignore this. This post should probably be a sticky somewhere for people new to the forum.

 

[j_j]

I've tried to explain this to so many people. And they either don't understand it or choose not to. I don't have the credibility that J_J has so too many people just ignore this. This post should probably be a sticky somewhere for people new to the forum.

I've no credibility with the fantasy-mongers. They express nothing but hatred and contempt for me.

 

[JohnA]

So are all these 'high-tap' people definitely deluding themselves?

And where would the threshold of audibility be? 100, 1000 taps?
How far the Nyquist-Shannon way do we need to go for any trained listener to be imposible to discern any further improvement?

It would be damn useful to have a decent blind test eh

 

[Chester]

So are all these 'high-tap' people definitely deluding themselves?

And where would the threshold of audibility be? 100, 1000 taps?
How far the Nyquist-Shannon way do we need to go for any trained listener to be imposible to discern any further improvement?

It would be damn useful to have a decent blind test eh

It’s a good question. I’m assuming there is no way to test the difference a high tap filter brings. Even if it was at what are considered inaudible levels, just being able to see some data that shows some impact it has (on something) would be interesting and a good start. But I’ve never seen anything so as I say, I’m assuming it’s not possible.

If the only way is controlled listening tests then I’m less interested, as I don’t listen to music in the same way a trained listener listens in a test……unless trained listeners perform their test while drinking lots of beer, slouching on the sofa

 

[j_j]

It’s a good question. I’m assuming there is no way to test the difference a high tap filter brings. Even if it was at what are considered inaudible levels, just being able to see some data that shows some impact it has (on something) would be interesting and a good start. But I’ve never seen anything so as I say, I’m assuming it’s not possible.

If the only way is controlled listening tests then I’m less interested, as I don’t listen to music in the same way a trained listener listens in a test……unless trained listeners perform their test while drinking lots of beer, slouching on the sofa

What's this got to with "hearing 300dB down"? If by high-tap filters you mean long FIR filters, that's a completely different issue than discussed here.

 

[Chester]

What's this got to with "hearing 300dB down"? If by high-tap filters you mean long FIR filters, that's a completely different issue than discussed here.

Yes, I was merely responding to the message before mine that mentioned them.

 

[j_j]

Yes, I was merely responding to the message before mine that mentioned them.

That should be its own thread. It's a much more complex issue than "length".

 

[Chester]

I've no credibility with the fantasy-mongers. They express nothing but hatred and contempt for me.

No more so than the comment above. Let’s move on and focus on the topic at hand.

 

[j_j]

No more so than the comment above. Let’s move on and focus on the topic at hand.

Sorry Chester, volunteers don't need abuse.

 

[Compact_D]

In the video, there is a reference to a pipe organ 200 feet away.
I assume the main depth info would come from timing and amplitude of reverberation. If we recorded it, should we not then play it back in the anechoic chamber?
Audiophile room is far from an anechoic chamber, so additional "short" reverberation from this small room would inevitably mess up the reverberation of the large cathedral and with it, the large distance perception.

If I am not totally wrong here, I do not understand how one could claim that DAC, however magical, can fix this.

 

[j_j]

One fun thing is the SPL in air has a lower limit due to thermal motion of the air itself, somewhere between -20 and -30 dB SPL.

https://physics.stackexchange.com/questions/110540/how-loud-is-the-thermal-motion-of-air-molecules

Be a little careful, that's for a 1000Hz bandwidth. For a 20000Hz bandwidth, considering the area of the eardrum, it's about +6dB SPL to +8.5dB SPL.

 

[JohnA]

What's this got to with "hearing 300dB down"? If by high-tap filters you mean long FIR filters, that's a completely different issue than discussed here.

I mentioned it because both claims come from the same crowd in the same context.
Maybe there is something into the claims, and not just plain old-fashioned expectation bias.
..I hope..

 

[j_j]

I mentioned it because both claims come from the same crowd in the same context.
Maybe there is something into the claims, and not just plain old-fashioned expectation bias.
..I hope..

Part of the problem is that "high-tap filters" is so completely off the wall by itself (aside from being ambiguous is it high gain, high tap weights, long length, high listener weight, what?) I'm not sure I want to start.

But in any case, there is a world of complexity in there that most assuredly does not simplify to 'long bad short good' or vice versa.