Vidcast:  https://www.instagram.com/p/CzpBtSBr3j-/

Stanford ER docs collaborating with University of Toronto bioengineers have developed speech analysis models for phones and also for smart speakers capable of detecting alcohol intoxication with 98% accuracy.  The researchers tested their algorithm on a small group of 18. Nearly three-quarters of the subjects were male with ages ranging from 21-62 years.  The participants were asked to read a tongue-twister after drinking an excess of weight-based alcohol over a 7 hour period.  The speech spectrographic patterns could predict a blood alcohol level over 0.08% and higher with a 98% sensitivity, a 97% specificity and positive predictive value, and a 98% negative predictive value.  It won’t be long before your car demands an error-free rendition of the flag pledge before sparking the ignition.

https://www.jsad.com/doi/10.15288/jsad.22-00375

#smartphone #smartspeaker #alcohol #breathalyzer #voicespectrum #intoxication