A computer programme has been used to predict how people are feeling when they type.
In a study participants were asked to type a particular phrase and the programme then estimated if they were happy, sad and so on.
Remarkably it was correct 70 per cent of the time, and the findings could lead to smarter artificial intelligence in the future.
The study published in the journal Behaviour and Information Technology was carried out by researchers at the Islamic University of Technology in Bangladesh, reports Live Science.
In the research 25 people ranging from 15 to 40 years old were asked to retype two paragraphs from Alice's Adventures in Wonderland by Lewis Carroll.
This allowed the programme an opportunity to understand how their emotions changed their typing style.
Moods tested were joy, fear, anger, sadness, disgust, shame and guilt.
The result was that the programme could identify a person's mood correctly 70 per cent of the time.
It was most successful in identifying joyfulness with an 87 per cent success rating, while sadness was the least accurate at 60 per cent.
The researchers noted that the participants were less likely to input data when they were in a bad mood, which may explain the inaccuracy of the latter result.