• artificial intelligence
• data visualization
• emotions
• 6D
• physiological signals
• stress
• relax
• spherison
• pillow

The thesis presents a proof of concept based on an experimental study to unveil underlying stress or apparent relaxation by combining 6D audio stimulation and artificial intelligence. The experiments involved recording various physiological signals while participants listened to a series of audio tracks purposely selected to elicit "stress" and "relaxation". The series comprised each audio track's 6D and stereo versions in random order. The collected signals included electroencephalogram (EEG), the electromyography (EMG) of facial muscles, electrodermal activity (EDA), photoplethysmogram (PPG), heart rate (HR), respiration rate (RR) and peripheral temperature (T). At the end of each audio track, each participant was asked to classify the track among "relaxing", " non-relaxing" and "I don't know”. Analyzing the signals, we tried to improve the understanding of the participants' emotional states. We generated a "graphical signature" of the signals associated with each participant-track pair using artificial intelligence. Then, we designed AI-driven data visualization techniques to make it easy to identify recurring patterns in participants who said they were relaxed but, in reality, they were not, and vice versa. A deep neural network classified the participant's emotion based on the graphical signature, also using frequency components. The results showed that a system implementing the proof of concept presented in this thesis could be promising to detect latent stress or apparent calmness. 6D audio stimulation could be crucial in unveiling potential states of apparent calm. This could help prevent chronic stress and control the underlying stress levels, for example in those affected by post-traumatic stress disorders.