Trombonist texted wife with voice recognition…while playing the trombone

Meet Paul The Trombonist, he is, as his Twitter name suggests a trombonist, and he says “I accidentally texted my wife with voice recognition…while playing the trombone”.

I accidentally texted my wife with voice recognition…while playing the trombone pic.twitter.com/tWCPSXbbrO

— Paul The Trombonist (@JazzTrombonist) November 21, 2017

If you’re wondering how this is even possible then he’s made a video to explain:

Hey everyone! Thanks to all for sharing. Many of you are asking how this was possible. I made this video to help explain. Warm regards https://t.co/cjfjZXWiT6

— Paul The Trombonist (@JazzTrombonist) November 23, 2017

And here’s the science bit

“It’s a recurrent neural network model that’s doing it. Recurrent neural networks take in time-sequenced acoustic information and are able to produce an output (written language) on the basis of that input. These programs are sensitive to anything that sounds like human phonemes. Your smartphone just happens to be designed with these recurrent neural network models embedded into it. That’s how Siri (if you’re an iphone user) can answer spoken questions, and how your phone can convert acoustic information into writing in general.” says @Dbozz71.

Source: Twitter/@JazzTrombonist