Relating FFT magnitude to dBSPL

[Catherine Julian]

Hello, please know I am a member of the public, not a student.

I use the Acceleration Spectrum feature of phyphox to monitor infrasound and low-frequency vibration (0.78 - 30 Hz at 5 - 72 dBSPL) happening constantly in my home and the surrounding geographic area. The source is unknown and I am collecting data to help me find where or what it's coming from.

I have some technical background in telephony and data networking but none in signal analysis or advanced mathematics. I hope nobody feels insulted that I ask here for help.

Before I found phyphox, I used another free app called Infrasound Detector that was much simpler -- in fact, too simple.  I read online about infrasound and wave mechanics, and want to collect better information about the 'signal' that is more accurate and more credible. I was happy to find phyphox! 

However, I don't know how to interpret the signal magnitude shown by the Fast Fourier Transform in phyphox [FFT Mag. (a.u.)], or ANY of the readings on the Raw Data page. 

The other app (Infrasound Detector) gave readings expressed as frequency (Hz) and intensity (dBSPL).  It would help me VERY MUCH to be able to understand the phyphox magnitude readings in terms of these units of measure.

Or, is there a different app that might suit my purpose better, instead of phyphox? I doubt it but thought I should ask.  I want to respect everybody's time on this forum as well as find an answer for my need.

I'm very grateful for whatever help or advice I receive on this topic.  Thank you in advance!

Catherine Julian
USA (Washington State)

[Jens Noriʇzsɔɥ]

Dear Catherine,

you are perhaps aware that the acceleration spectrum does not help you that much with infrasound matters. I am also a bit unsure if smart device mics (the electronics part) are sensitive to infrasound at all, they would miss most of the range at least…

If the infrasound levels are high enough to cause vibrations, these could be seen in the acceleration spectrum. “dbSPL” is an acoustics measure (*S*ound *P*ressure *L*evel) that makes no sense for kinematics. Acceleration is given in m/s² (velocity change per time interval).

The Fourier transform of the recorded acceleration gives you an impression of the relative intensities of frequencies, that's why it is given in arbitrary units (a.u.). If you put your smartphone on a washing machine rotating at 1200 rpm, you would see a sharp peak at 20 Hz, for instance.

[Catherine Julian]

Dear Catherine,

you are perhaps aware that the acceleration spectrum does not help you that much with infrasound matters. I am also a bit unsure if smart device mics (the electronics part) are sensitive to infrasound at all, they would miss most of the range at least…


Hello Jens, thanks very much for your reply!  Your input helped me understand the FFT readings much better.  You are correct, it is not a source of what I want to know. Too bad for me.

Definitions can get unclear and contradictory with mechanical wave frequencies between about 1 Hz and about 50-60 Hz -- 'sound' is arbitrarily said to be what 'most' humans can 'hear'  (20-20000 Hz)but hearing isn't the only way humans perceive vibrations, and the nerves of the hearing center sends messages to many parts of the brain. I talk about frequencies only because I know nothing about the kinetic perspective and sciences. Sorry!

I've read that infrasound (2-20 Hz) acts like seismic ground waves so is appropriately detected via accelerometer in a smartphone, not the microphone. Also it causes resonance in building materials and human tissues such as vertebrae, lungs, eyes, viscera, teeth, sinuses, etc., which are mechanical waves traveling thu various coupled media (ground, structure materials, air human tissues).  Sound or not sound? Opinions disagree between fields.
 
If the infrasound levels are high enough to cause vibrations, these could be seen in the acceleration spectrum. “dbSPL” is an acoustics measure (*S*ound *P*ressure *L*evel) that makes no sense for kinematics. Acceleration is given in m/s² (velocity change per time interval)


Vibrations are tangible, and can be seen on surfaces of beverages (coffee, tea, water) by watching reflections.


The Fourier transform of the recorded acceleration gives you an impression of the relative intensities of frequencies, that's why it is given in arbitrary units (a.u.). If you put your smartphone on a washing machine rotating at 1200 rpm, you would see a sharp peak at 20 Hz, for instance.


This helps very much! I understand better now! Many Thanks!!

Catherine Julian

[Jens Noriʇzsɔɥ]

You are right, I have not been fully precise in my wording on sound. With smart devices we could just deal with airborne sound via microphone (and speaker) and that's what my explanations were targeted at.

There is so much information missing on sound coupling, frequency dependent reference values for dBSPL, and others that m/s² remains the only available scientific measure that the accelerometer could provide.