Chest expansion reconstruction from respiration sound by using artificial neural networks

Affective computing is a growing area in which researchers are focusing on the recognition of emotions through the analysis of biomedical signals. Emotion recognition is useful when it is done during real life activities; this is possible only by the use of devices that can be easily worn by the subject and that do not affect his/her activities. In this work, we present a way to reconstruct the chest expansion signal, usually measured by an uncomfortable belt around the chest, from the analysis of respiration sound gathered with a microphone placed on the upper part of the neck. We will show that it is possible to reconstruct the respiration spectrum with an error lower than 0.06 Hz in the frequency that characterizes it.