Mandarin Emotional Speech Rhythm Analysis

Together with the rapid technological developments in human-computer interaction in recent years, emotional information processing within speech signals has becomed one of the focal points in phonetics research. This paper primarily deals with analysing the prosody of emotional speech.In this study, uses the method of combined the results of acoustic analysis and perceptual experiment, of comparative analysis and statistical analysis, based on the establishment of the corresponding emotional speech database and more in-depth study and discussion in emotional speech prosody.We found through research:Happy, anger, sadness, surprised and neutral tone of voice characteristics, there are some significant differences in the prosody, such as duration, the maximum fundamental frequency, minimum fundamental frequency, accented tone location, duration, etc. Specific performance is as follows:In the results of stress perception:Emotional stress the most frequent in anger, sadness the lowest frequent of emotional stress. More than happy in the end of the sentence stress; Anger stress more in the middle of sentences; Sadness more stress in the middle and end of sentence; Surprise stress more for the end of the sentence.In fundamental frequency (F0) of sentences and tone:happy, anger, sadness, surprised, neutral tone domain width relationship:anger> surprised> happy> neutral> sadness; Happy's F0 smoother than anger in transition between high and low values; happy is the high-profile domain-based narrow field-based; Anger's F0 are distributed in the high frequency band, and high-profile domain-based width field-based; Sadness little change in the fundamental frequency and are located in the low frequency band, and low-profile domain-based narrow field-based, the position of the syllable words more to maintain the tone type; Surprised rise in fundamental frequency end of the sentence,and high-profile domain-based width field-based.In F0 performance of the stressed syllable sentence:anger's F0 range in stress position widened the most, followed by surprised. Sad and neutral widened slightly. Happy's F0 range in stress position widened larger than sadness and neutral.In duration feature:happy, angry, sadness, surprised, neutral speech speed is: anger> happy> neutral> surprised> sadness; Anger stress duration accounted for the entire length of sentence length of time the largest proportion. Sadness's stress duration accounted for the entire length of sentence length of time the smallest proportion. Surprised the biggest increase in stress syllable durationIn pause:pause the least number of happy, shortest in the total length of the pause; Sadness pause most often, longest in the total length of the pause.In summation, the new concepts introduced by this paper can be displayed as:In order to investigate the quality of the text, a perceptual experiment was performed, distinguishing the type, focus and extent of the emotion, creating a more comprehensive text corpus, and thereby being able to do deeper research.This paper uses the combined results of the acoustic analysis and the perceptual experiment to analyse and describe the characteristic prosody of emotion speech, enhance the motivation of the analysis.