語料參數與聽者在語音情緒辨識之關聯性研究

人與人互動中，情緒是很重要的，人的情緒可以透過臉部表情，手勢和語句表現出來，讓聽者可以輕易地了解語者當下的感受。在這個研究中，我們針對一個新建立的語料庫研究語者所要表達之情緒和聽者所感受之情緒一致性的關聯，這個與料庫包含六種基本情緒和八種複合情緒，分別有12,749及6,387句。為了從這個語料庫分析語者表達情緒與標記者感受情緒間的差異性，我們使用統計的方法包含平均數，中位數和標準差。我們也比較不同語音持續時間對標記者感受情緒的差異，同時也和情緒辨識系統的結果作比較。從實驗結果可以得知，女生比起男生對情緒有更好的表達能力，但在情緒感受上卻有較大的變異性。此外，我們也發現語音持續時間越長，得到的情緒辨識結果越好。

關鍵字

語音情緒辨識；權重離散第k位最接近的鄰居；持續時間偵測；中文語料庫；標記

並列摘要

Emotion plays an important role in interaction among humans. It can let listeners easily understand the feeling of speakers expressed in facial expression, gesture and/or speech. In this research, we investigate the relation of the intended emotion expressed by speakers and the perceived emotion by the listeners of a newly constructed corpus. In this corpus, the expressed emotion includes six basic emotions with 12,749 utterances and eight derived emotions with 6,387 utterances. We use statistical method such as mean, median and standard deviation, to find the consistency of the emotion expressed by speakers and the perceived emotion by listeners. Furthermore, we also compare the results from annotation of utterances of different duration with the results from emotion recognition system. The results indicate that females have better emotion expression ability than males. But they have larger variance in emotional perception. Another important finding is that the recognition rate has a tendency of increase when the duration of utterance getting longer.

並列關鍵字

speech emotion recognition ； weighted discrete-KNN ； duration detection ； chinese mandarin ； tendency

參考文獻

[1] A. Ortony and T.J. Turner, “What's Basic about Basic Emotions,” Psychological Review, pp. 315-331, 1990

[2] D. Canamero and J. Fredslund, “I Show You How I Like You: Human-Robot Interaction through Emotional Expression and Tactile Stimulation,” Retrieved 2010-05-20, http://www.daimi.au.dk/~chili/feelix/feelix.html

[5] J.H. Yeh, Emotion Recognition from Mandarin Speech Signals, Master Thesis, Tatung University, 2004

[7] R. Cowie and E. Douglas-Cowie, “Automatic statistical analysis of the signal and prosodic signs of emotion in speech,” Proc. 4th International Conference on Spoken Language Processing, pp. 1989-1992, 1996

[8] M.W. Bhatti, Y. Wang, and L. Guan, “A neural network approach for human emotion recognition in speech,” Proc. the International Symposium on Circuits and Systems, vol. 2, pp. 181-184, May 2004

國際替代計量

語料參數與聽者在語音情緒辨識之關聯性研究

未授權

主題瀏覽