Biometrics for Emotion Detection (BED): Exploring the Combination of Speech and ECG

Abstract

The paradigm Biometrics for Emotion Detection (BED) is introduced, which enables unobtrusive emotion recognition, taking into account varying environments. It uses the electrocardiogram (ECG) and speech, as a powerful but rarely used combination to unravel people’s emotions. BED was applied in two environments (i.e., office and home-like) in which 40 people watched 6 film scenes. It is shown that both heart rate variability (derived from the ECG) and, when people’s gender is taken into account, the standard deviation of the fundamental frequency of speech indicate people’s experienced emotions. As such, these measures validate each other. Moreover, it is found that people’s environment can indeed of influence experienced emotions. These results indicate that BED might become an important paradigm for unobtrusive emotion detection.