A Comparison of KNN Based Classifiers for Detecting Emotion from Mandarin Speech

Abstract

Humans communicate through speech, movement, hand gestures and facial expressions. We express our emotions in speech by the words that we use and intonation of the voice. Whereas research about automated recognition of emotions in facial expressions is now very rich, research dealing with the speech modality has only been active for very few years and is almost for English. In this paper, we presented a comparison of three KNN based classification algorithms for detecting emotion from Mandarin speech. The results show that the proposed weighted D-KNN outperforms the other two classification techniques: 13.1% improvement for traditional KNN and 7.4% improvement for M-KNN. The highest recognition rate (79.31%) is obtained with weighted D-KNN using Fibonacci series.