Recommend sofware with high-precision FFT analysis with resolution to -384dB?

[Uncle Freddie]

Hi,

Can anyone recommend Mac and/or PC audio analysis software packages that -- and here's the tricky part -- offer high-precision FFT analysis with resolution all the way down to -384dB? That would be 64 bits of resolution.

Thanks,
Fred

[Folderol]

Good Lord!
What on earth for?

[Hugh Robjohns]

that was going to be my question, too!

[Uncle Freddie]

I'm looking for this depth of resolution for test and measurement of truncation noise and other distortions.

In macOS, I have used Aurchitect's Spectre software for this, eventually switching to their Aura package. Both are considerably buggy. On top of that, in mid-May the company went out of business over ownership lawsuits. They even wiped their website (http://aurchitect.com) completely off the air.

I recently began using SPL Hawkeye, but it only goes down to -220dB (I have filed an enhancement request to increase that to -384dB)

Surely there must be software out there, Mac or PC, perhaps aimed at the test and measurement audience, that can do this?

Thanks,
Fred

[Martin Walker]

I must admit I'm still a bit in the dark - highly regarded plug-in developers have stated that in most cases for audio work you can declare anything further down than -125dB below the signal as inaudible

Martin

[ef37a]

I must admit I'm still a bit in the dark - highly regarded plug-in developers have stated that in most cases for audio work you can declare anything further down than -125dB below the signal as inaudible

Martin

Quite. Nothing is that quiet except perhaps liquid Helium?

Dave.

[Sam Spoons]

Is there a theoretical measurement for 'absolute silence'?

[Martin Walker]

I'm beginning to suspect that Uncle Freddie may be exploring mathematical concepts/algorithms, rather than real-world audio.

However, I'd still be surprised if any signal that low would be worth exploring, apart from in a cutting edge lab environment.

Fascinated to hear more though!

Martin

[ef37a]

Is there a theoretical measurement for 'absolute silence'?

Well, as per my "He" at absolute zero, by definition energy is zero* so that has to be zero noise?

*I understand there can still be electron spin but can you "hear" electrons? BTW "I understand" is just a turn of phrase. I don't!

Dave.

[Logarhythm]

Is there a theoretical measurement for 'absolute silence'?

Well, as per my "He" at absolute zero, by definition energy is zero* so that has to be zero noise?

*I understand there can still be electron spin but can you "hear" electrons? BTW "I understand" is just a turn of phrase. I don't!

Dave.

There is no absolute zero, Dave
Well there is in terms of 0 Kelvin, but quantum mech prohibits this being a truly zero energy state and it's perhaps better viewed as the lowest energy state - thus there will still be "noise" of sorts, and indeed noise is perhaps an appropriate adjective for e.g. quantum vacuum fluctuations.

And I'm also really curious to know what Uncle Freddie is up to!

[Drew Stephenson]

It's amazing the things you learn on here.

[ConcertinaChap]

There is no absolute zero, Dave
Well there is in terms of 0 Kelvin, but quantum mech prohibits this being a truly zero energy state and it's perhaps better viewed as the lowest energy state

That's true. Interestingly from a classical perspective as well you can't reach absolute zero because it's one of these boundary conditions, like the speed of light, that require greater and greater amounts of energy the closer you approach them and would require infinite energy to actually achieve.

And I'm also really curious to know what Uncle Freddie is up to!

CC

[Wonks]

Can't be for audio, as the electronic noise of the input system will create a noise floor way above the lowest resolution of the analyser. That limits the resolution of a standard 24-bit A/D audio converter to maybe 21 bits, 22 at best. So even if you used a 64-bit A/D converter set up with it's upper limit at around 150dB, I'd guess you'd be wasting at least 30 of those bits as they'd be below the electronics noise floor.

[Uncle Freddie]

I would be delighted to tell you all what I am up to, but I cannot at the moment. Soon, though.

In case I was not clear earlier: Aurchitect's Aura indeed has FFT resolution down to -384dBFS. What I am asking for is not unheard of.

I still use it, or rather I try desperately to use it, because I am fed up with its bugs. Which by the way will never be fixed, now that the company has gone belly-up.

Aura is Mac software, but I am entirely open to a Windows utility for this.

Thanks,
Fred

[James Perrett]

For serious analysis work most people would be thinking of Matlab or the free alternatives like Scilab or Gnu Octave.

[DC-Choppah]

Mathematica supports data types with arbitrarily large precision. You can do an FFT on data with as high a precision as you want.

X:=N[{0,0,0,1,1,2,1,1,0,0},100]

Defines a waveform stored numerically with a data type having 100 significant digits.

Any process you apply will be done with that precision such as an FFT.

Fourier[X]