A MEASUREMENT OF HUMAN VOCAL EMOTION
USING FUZZY CONTROL

ABSTRACT

In this paper, we propose the system that measures the degree of the specific emotions on speech. The past studies investigating the relationship between physical parameters and emotions are based on the assumption that they have linearlity and are able to be treated separately. In practice, however, man captures some vocal physical changes synthetically and then makes evaluation with plural emotional elements at a time. Fuzzy control seems to be suitable engineering method to modeling such an object. For the input of physical changes from the neutral state on speech, each degree of anger, sorrow, amazed and fear communicated by the speech is output. The system is independent on personalities and words because it treates emotions perceived by listeners and physical parameters are relative to neutral speech. Fuzzy inference rules and the shape of the input membership functions are optimized dependent on speeches that include specific emotion. It is shown that the system is equivalent to subjects with respect to measurement of emotional content on speech and the validity of the model is confirmed.

REFERENCES

1. Tsuyoshi MORIYAMA, Satoshi MACHII, Hwang JUNGYUN, Shinji OZAWA
   ``A Measurement of Human Vocal Emotion using Fuzzy Control'',
   IEICE, A-304(1994-03)
2. Tsuyoshi MORIYAMA, Shinji OZAWA
   ``A Measurement of Human Vocal Emotion using Fuzzy Control'',
   Journal of IEICE D-II, Vol.J82-D-II, No.10, pp.1710-1720, October 1999.