The Natural Language Processing Research Group covers a wide range of investigation areas within the framework of the Natural Language Processing. The Sounds of Science had two collaborators from TALN, who work with prosody and gesture analysis to predict orator gesture by the speech content and expression.

The prosody consists on analyzing the oral expression features such as the expression, tone, etc. The TALN group is working on techniques to generate realistic speech audios from text files, using prosody to generate a voice which considers intonation, expression, and other important variables of human speech. Here are some examples of synthetic speech generated by these methods.

For Natural Language Processing tasks, is very common to use mechanical keyboards, which provide a high commodity and precision for typing.

The other project that we covered when we visited TALN was about gesture analysis. This scientific work consists on estimating orator gestures by the text and talking expression. Many people, especially when talking in public, move their arms and hands to complement the explanation. These are the gestures that want to be estimated in this project. Most of these gestures, produce characteristic sounds and it’s interesting to take them into account.