Research Picks Extra – October 2015

Vocal stress detection

Our voices change when we are under stress, for instance, people often talk higher and faster. Automating the detection of stress or fatigue in users of voice-controlled systems could be used to improve safety and security. Researchers in India have carried out an analysis of emotionally charged speech as well as neutral voices to tease out the details. They report that the spectra of relaxed and stressed speech are different, with the fundamental frequency of the vocal tract elevated in someone stressed. Moreover, characteristics of the voice known as formants are different in subtle ways that are perhaps obvious to another person listening to the speech but might not be embedded in a digital system to automatically recognise when a person is stressed.

Sondhi, S., Khan, M., Vijay, R., Salhan, A.K. and Chouhan, S. (2015) ‘Acoustic analysis of speech under stress’, Int. J. Bioinformatics Research and Applications, Vol. 11, No. 5, pp.417–432.

 

Keeping us interested

How long does a new update on a social network maintain its users’ interest. US researchers have investigated four popular sites – Reddit, 4chan, Flickr, and YouTube – to ascertain the lifespan and popularity of content. Data mining and statistical techniques have allowed them to build a model of content longevity that is 95% accurate in predicting lifespan and could be of use to educators and marketers looking to utilise online social networks to get their message home more effectively.

Gibbons, J.W. and Agah, A. (2015) ‘Modelling content lifespan in online social networks using data mining’, Int. J. Web Based Communities, Vol. 11, Nos. 3/4, pp.234–263.

 

The leapfrog shuffle

Researchers in China have used a shuffled frog leaping optimisation algorithm to help them detect communities in a social network. Given that there are hundreds of millions, if not billions, of people using online social networks, it is important from the social science perspective as well as from the business and political points of view to be able to simplify the complexity of these vast networks by finding communities within the wider network in which members have shared interests, affiliations and political persuasions. The team’s modified leapfrog algorithm gives more precise results than other algorithms that have been used on the same test data for classifying users according to community membership.

Wang, T., Zhao, X. and Zhou, Y. (2015) ‘Community detection in social network using shuffled frog-leaping optimisation’, Int. J. Security and Networks, Vol. 10, No. 4, pp.222–227.

 

A gesture token

Scientists in Taiwan have developed a real-time hand gesture recognition system for the retrieval of information from the internet that sidesteps endless mouse clicks and keyboard shortcuts. The system uses face recognition to identify the user of a given computer with a set of gestures taught to the computer by that user for carrying out particular tasks. A user might set an index finger pointing up to signify they want to read a particular site, a fist might give them the weather forecast, a thumbs-up take them to their favourite online comic etc.

Tsai, J.C., Chang, S-M., Yen, S-H., Li, K-C., Chen, Y-H. and Shih, T.K. (2015) ‘A real-time hand gesture recognition system for daily information retrieval from the internet’, Int. J. Computational Science and Engineering, Vol. 11, No. 2, pp.105–113.

Author: David Bradley

Award-winning, freelance science writer based in Cambridge, England.