Understanding FFT in Audio Measurements

[Plcamp]

I am wondering when Amir will get one of these?

 

[amirm]

Wow, had not heard of that machine. Remarkable. Thanks for posting.

 

[Plcamp]

Wow, had not heard of that machine. Remarkable. Thanks for posting.

I cringe at the thought of people who figured stuff out eons ago, but lacked the tools we have now. This example is so cool...and better than any I have seen will cause folks to actually understand what is going on!

Interestingly, the govt of Canada was one of the few purchasers, wonder where that unit is now...in the bowels of the NRC here in Ottawa I bet.

 

[amirm]

I hear you. My brother bought a bunch of ASR 33 teletype machines and I remember always marveling at the complexity of those machines. It seemed to have a million moving parts! Naturally, no matter how old or odd, someone is restoring them!

 

[Plcamp]

Heh, I always thought as an engineer that the road to hell is paved with mechanical design constraints, and congratulated myself secretly for not getting involved in the photocopier design business.

 

[SIY]

My first FFT system:
http://versci.com/images/nic7199.jpg

 

[pma]

My first FFT system:
http://versci.com/images/nic7199.jpg

Our 1982 project of ADC waveform digitizer + memory

 

[pozz]

My first FFT system:
http://versci.com/images/nic7199.jpg

What was it like working with it?

 

[SIY]

What was it like working with it?

Slow.

By the time I got hired at Nicolet, they had reduced the size to merely 18 cubic feet or so.

 

[TheBatsEar]

Thanks for that video, very interesting. Turns out my digital scope, a Siglent SDS1202X-E, has simple FFT as well.

 

[krabapple]

I enjoyed the video. I've used FFT for a different purpose: in Audition to do comparative frequency analysis of different remasters of music releases. But I'm still just guessing what window function and number of FFT bins to use. It appears to me that to 'resolve' the bass frequency content accurately requires a more FFT bins. Why is that? And what impact does that have on accuracy or ease of visual interpretation in the treble?

in short: Any best practices recommendations for doing this sort of analysis?

 

[Blumlein 88]

If you are using an FFT on a 48 khz file you have 24,000 hz to work with. If it is a 1024 bin FFT it will divide the 24,000 into 512 bins or each bin will cover about 471 hz. So all the bass will be in that one 471 hz wide bin which is not much resolution. You will have one value for each bin by what it contains.

Now if you use a 32k or 32,768 bin FFT it will divide 24,000 by 16,384 bins or 1.46 hz per bin which is pretty good resolution even in lower frequencies.

For windowing read this, as it explains it likely better than I would.
https://www.egr.msu.edu/classes/me451/me451_labs/Fall_2013/Understanding_FFT_Windows.pdf

 

[JG2020]

Excellent! I've seen this table before but could not find it again.

from here - https://download.ni.com/evaluation/pxi/Understanding FFTs and Windowing.pdf

 

[krabapple]

If you are using an FFT on a 48 khz file you have 24,000 hz to work with. If it is a 1024 bin FFT it will divide the 24,000 into 512 bins or each bin will cover about 471 hz. So all the bass will be in that one 471 hz wide bin which is not much resolution. You will have one value for each bin by what it contains.

Now if you use a 32k or 32,768 bin FFT it will divide 24,000 by 16,384 bins or 1.46 hz per bin which is pretty good resolution even in lower frequencies.

For windowing read this, as it explains it likely better than I would.
https://www.egr.msu.edu/classes/me451/me451_labs/Fall_2013/Understanding_FFT_Windows.pdf

The three window functions I see used most often for music frequency/amplitude analysis are Blackman(-Harris), Hamming, and Hanning ('Hanning' , aiui, is a mistaken reference to the 'Hann' window)

I presume music approximates 'random' signal type, which all three of thase are 'best' for in that table
For frequency resolution, Blackman is 'poor' the other two 'good'
For Spectral Leakage, they are best, good, and fair, respectively
For Amplitude accuracy, they are good, fair, and fair

Since mastering comparisons are basically amplitude comparisons, I would guess Blackman is best to use of? Though its frequency resolution is poor? (Then again...Blackman-Harris isn't the same as Blackman...)

 

[mansr]

I think Dolph-Chebyshev deserves a mention.

 

[SIY]

I think this should not be windowed.

 

[telecode101]

Hello, thank you so much for the videos and post. I am going through them just trying to learn and understand if a difference of -85 vs -95 of a noise floor is worth $700 .

 

[Ashley Salmond]

FFTs are the fundamental to many measurements I show in my reviews. You can search for it online but I am not sure you will find any easy to understand explanation of it that applies to the way we use them in audio measurements. So I thought I do a walk-through using the Audio Precision of what it means, and how its parameters can drastically change what it displays.

As I asked the video, please let me know if you like these kinds of technical tutorials. To be sure they are not as much "fun" to watch as audio debunking ones. So like to know if you want to see more of them or not. And whether you could follow this one which is probably one of the most difficult concepts we have in audio measurements.

While I have not done a full text write up on this, the concept is included in my tutorial on digital audio measurements: https://www.audiosciencereview.com/...derstanding-digital-audio-measurements.10523/

Hi Amirm, I like the tutorials, but without drawing and diagrams and ,say a little more layman approach I think I would understand them better. It's not you but more from my position as I am not an academically inclined but do enjoy science just the same

 

[Human Bass]

Very cool vídeo on the history of FFT

 

[SIY]

For anyone in western NY, I'll be teaching a course on Fourier transforms and their use in measurement here at Alfred University starting in mid-December. PM me if you want to audit.