This is probably more 'voice' than breathing, but when I'm in the toilet cubicle at work I try to identify anyone who may be next door by the sound of their breathing.

I rarely get to confirm whether I'm right or wrong, but everyone sounds slightly different.

> When 42 of the participants came back to the laboratory weeks, months and even two years later, to take part in another 24-hour measurement, the trained algorithm could identify them from their breath patterns. Data from periods when the participants were awake gave more accurate results than did those from sleeping periods, but when the researchers used a 100-parameter characterization of a full data set instead of one using 24 parameters, they could pick individuals out with 96.8% accuracy.

The correctly identified .968x42=40.696 of the participants.

Also any standard-ish physical activity that comes with instructions usually includes breathing in those instructions. So I would expect results to vary substantially depending on where they found the study participants.

Including asymmetry between the nostrils brings in physiological factors other than breathing, i.e. sinuses, etc..

Still, I can see it. My wife and I are probably equally fit, but she breathes much faster than I do. I also notice that I sometimes don't take a breath (or feel any need to) for several seconds, if I'm being sedentary.

Previous: https://news.ycombinator.com/item?id=44262799

I have noticed that I need so much fresh air while sleeping that it is not very comfortable for me to sleep with another person. I can not say anything about breath patterns but I suspect that O2 consumption has to be among those patterns.

This is brilliant. Could definitely use this for a very low-cost diagnostic tool. It's like reading someone's pulse, but through their breath.

You are now breathing manually.