2015 paper, https://dl.acm.org/doi/abs/10.1145/2789168.2790109

> We implemented the WiKey system using a TP-Link TL-WR1043ND WiFi router and a Lenovo X200 laptop. WiKey achieves more than 97.5% detection rate for detecting the keystroke and 96.4% recognition accuracy for classifying single keys. In real-world experiments, WiKey can recognize keystrokes in a continuously typed sentence with an accuracy of 93.5%.

More papers on keystroke recognition: https://scholar.google.com/scholar?hl=en&as_sdt=5%2C31&sciod...

60Ghz mmWave radar (coming to Wi-Fi via 802.11ay) has higher resolution, shorter range and low ability to penetrate building walls, but can pass through unshielded windows: https://news.ycombinator.com/item?id=30172647.

Thankfully I use an extreme custom keyboard layout so I wish them luck.

That's about as hard as decrypting substitution cipher. Just think about the problem for a second, you will figure out how to trivially go around it.

Well I think with enough time you could just use the key press model to try different configurations until you got one that gave intelligible results. The process could be easily automated if the resolution is enough to support accuracy in the 90th percentile

That’s just as secure as ROT13

Do you also have your number keys remapped non-linearly?