2015/10/22 by Şefik Emre Eskimez, Kenneth Imade, Eskimez, Sefik Emre +9 · 1 citation
Psychology · Computer Science · #Emotion and Mood Recognition #Sentiment Analysis and Opinion Mining
paper · pdf · doi:10.48550/arxiv.1510.06769
The fact that emotions play a vital role in social interactions, along with\nthe demand for novel human-computer interaction applications, have led to the\ndevelopment of a number of automatic emotion classification systems. However,\nit is still debatable whether the performance of such systems can compare with\nhuman coders. To address this issue, in this study, we present a comprehensive\ncomparison in a speech-based emotion classification task between 138 Amazon\nMechanical Turk workers (Turkers) and a state-of-the-art automatic computer\nsystem. The comparison includes classifying speech utterances into six emotions\n(happy, neutral, sad, anger, disgust and fear), into three arousal classes\n(active, passive, and neutral), and into three valence classes (positive,\nnegative, and neutral). The results show that the computer system outperforms\nthe naive Turkers in almost all cases. Furthermore, the computer system can\nincrease the classification accuracy by rejecting to classify utterances for\nwhich it is not confident, while the Turkers do not show a significantly higher\nclassification accuracy on their confident utterances versus unconfident ones.\n