I have been interested in delay-line memory for a while. It occurred to me that I could probably have my computer speaker play tones, pick them up with my microphone, and feed it back, thus storing bits "in the air".

A quick chat with an agent got it up and running. See attached github repo.

In my setup, a lot of the data is actually stored in the latency of the audio processing. I moved the mic closer to the speaker due to the bad acoustics of my room.

This is not at all practical, but it was fun.