Sigh. I've coded too much this week. My brain just immediately went to: comment out half the city at a time and binary search the source. Guess you can't really do that.
I wonder if the hum persists during a power outage.
"The University of Windsor report said the hum’s likely source was blast furnace operations on Zug Island" ... which is over the border in the United States.
Cutting off the power to that island would be a good thing to try, but apparently the US is uncooperative.
I'd guess amplitude is massively affected by microphone sensitivity. This is not easy to measure and might depend on the direction.
Consider:
* The amount of walls between a mic and the outside.
* The absorbance/reflection of nearby materials, including their shape (flat surface will reflect sounds amplifying certain directions).
* The clutter between the source and receiver. Sound carries over water, not over cities or forests.
* Similarly, elevation of the receiver. Higher means less effect of clutter, and obviously, a hill in between will be an issue.
* Wind direction. Wind will blow the sound away.
It might still work with enough measurements with the same microphone, but that is not a given.
A decent backup plan would be measuring the phase difference. This is still frustrated by wind, but doesn't depend on the senstivity of the microphones.
It does require synchronization of the recordings, which might also be difficult.
It does require synchronization of the recordings, which might also be difficult.
Could you record with a multiple "microphone connected to a Raspberry Pi" setup, with the time synchronized using something like the NIST time signal[1] using an RTL-SDR, or a cellular signal, or GPS, or maybe using NTP?
Yes, that seems like the best approach. At 20Hz waves, ntp would probably be good enough.
The real issue I can see is to get that time-data into the sound data.
Maybe you could time-stamp the beginning of a recording, and start a new recording every minute (to compensate clock drift).
That might work, biggest issue I see is potential variable latency between 'starting a recording' and the actual first recording happening.
I wonder if the hum persists during a power outage.