This past week, my main goal was to verify a stable multi-mic connection among the Adafruit I2S microphones. In order to test this without a Jetson Nano (since we were not sure when placing our first batch of orders whether we could afford getting every single team member a Nano), I did some research to find that the mics would work with an Arduino Zero, a smaller and significantly cheaper chip with which I could test them. I then wired up the Arduino and two of the mics, successfully receiving input from both.

From there, I learned more about stereo arrangements and tested how the two mics could pick up sound from different positions relative to each other. I also learned about acoustic location and multiple microphone arrays and realized that that would be more challenging and time-consuming than originally planned. I’ve broken down the steps more and adjusted the Gantt chart accordingly. Although it will take more time than anticipated, this shouldn’t throw us too badly off schedule, since we’ve also nixed the voice recognition part of our project after receiving feedback on our proposal and performing further research that confirmed the exceeding complexity of this task.

Going off of the new and more incremental goals I’ve laid out for myself, this coming week, I intend to write and test an acoustic location algorithm for the mics. My preliminary idea for an algorithm is to divide each microphone’s sampling dataset into partitions, find the time of local maximum of the amplitude for each partition, calculate the time difference between those times among the mics for each partition, and average the time differences between each pair of mics. For every pair, I’ll use that time difference and the metrics of the mics’ positions relative to each other to calculate the angle of the source, and finally, average all those calculations to get the best approximation of where the source is.