|
The purpose of our research is to enable computer develop more natural man-machine conversation. Therefore, it has to understand humans’ sentiments and learn from the nuance and context of every interaction. It’ll be rather difficult to determine when we take interactions literally, as the sentiment change mainly reflects in intonation, speed, rhythm etc. Accordingly, this research deals with the features of voice signal’s rhythms and analyzes the distribution of these features in various emotional states, so that it can provide important basis for the understanding of muti-emotional states in voice signals.
|
