And just sticking to counting, a not exceptionally well-trained ear could already count how many letters you typed and if you pressed backspace (at least with the double-width backspace, sound is definitely different)
Yeah I recall that there was an attack researchers demonstrated years back of using recordings of typing with an AI model to predict the typed text with some accuracy. Something to do with the timings of letter pairings, among other things.
93% - 95% accuracy and it wasn't even a good quality recording
> When trained on keystrokes recorded by a nearby phone, the classifier achieved an accuracy of 95%, the highest accuracy seen without the use of a language model. When trained on keystrokes recorded using the video-conferencing software Zoom, an accuracy of 93% was achieved, a new best for the medium.
Notably, I believe this has to be tuned to each specific environment. The acoustics of your keyboard are going to be different from mine. Which is not much of a barrier, given a long enough session where you can presumably record them typing non password-y things.
