sound level measurements
edsard boelen, 16 jan 2020
The esp32 is a dual core 32 bits controller with a FPU this is powerful enough to do real time spectrum analyzing.
This is fast enough for accurate noise measurements. The inmp441 is a digital microphone with i2s interface.
inmp441
the inmp441 is a mems microphone according to the datasheet it sends its audio signal at a datarate of 2.7 Mbit/s.
at least 50 measurements are made per second which will include all frequencies up to 10KHz.
hardware
software
The software has the following features:
- esp32 devkit
- platformio vscode libraries
- wifi details in eeprom
- locked https certificate
- async webserver
- OTA (over the air) updates
- i2s driver
- arduino FFT library for spectrum calculations
- direct noise level leds
- direct spectrum print on serial out
- 3 sec full average calculations
- spectrum average calculation and logging every minute
- generate and post json data every 5 minutes
logging and analyzing
for now, every 3 seconds the average total sound level is being logged. every minute the average spectrum is being logged on 8 Octaves.
This is sent as a json message every 5 minutes.
{
"date":"2023-06-25 00:04:08",
"spectrums":[[16.6,19.85,20.16,16.81,11.34,9.78,8.56,10.28,13.22],[13.38,17.59,19.2,16.51,11.4,8.5,8.08,10.12,12.72],[14.03,18.8,20.07,15.47,10.59,8.93,8.52,10.29,13.64],[13.71,17.72,17.67,14.01,9.51,8.3,6.98,9.77,12.81]],
"noise":[17.51,16.94,17.4,17.82,17.06,17.49,17,17.22,17.87,17.05,17.04,17.2,17.24,17.31,17.05,17.33,17.33,18.1,16.93,17.34,17.46,17.49,17.33,17.73,17.59,17.81,17.94,18.12,17.32,17.65,17.62,17.73,17.41,16.75,17.32,17.05,16.43,16.76,16.45,17.16,17.01,16.56,16.74,17.07,17.75,16.93,17.87,17.74,17.64,17.68,17.53,17.21,17,17.74,17.9,18.1,18.14,17.85,17.78,17.79,17.88,17.21,17.79,17.56,17.55,17.57,17.69,17.64,17.67,17.53,17.13,16.95,17.57,17.35,17.26,16.64,16.61,17.51,17.12,17.76,17.41,17.56,17.61,17.78,17.8,17.67,17.47,17.69,17.38,17.65,17.59,17.39,17.79,17.49,17.43,17.87,18.03,17.8,17.68]
}
after a full day of recordings, the following carpet chart / heatmap can be generated.
for this I used python with the plotly library.
