Emotion Recognition and Evaluation of Mandarin Speech Using Weighted D-KNN Classification
碩士 === 大同大學 === 資訊工程學系(所) === 93 === In this thesis, we proposed a weighted discrete K-nearest neighbor classification algorithm for detecting emotion from Mandarin speech. The Mandarin emotional speech database used contains five emotions including anger, happiness, sadness, boredom and neutral emo...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2005
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/95339126675789925087 |
| Summary: | 碩士 === 大同大學 === 資訊工程學系(所) === 93 === In this thesis, we proposed a weighted discrete K-nearest neighbor classification algorithm for detecting emotion from Mandarin speech. The Mandarin emotional speech database used contains five emotions including anger, happiness, sadness, boredom and neutral emotion. The extracted acoustic features include Mel-Frequency Cepstral Coefficients (MFCC) and Linear Prediction Cepstral Coefficients (LPCC). The best recognition rate achieved was 79.55% obtained with weighted discrete K-nearest neighbor classification.
In addition, we also proposed an emotion evaluation method using weighted discrete K-nearest neighbor classification. We designed an emotion radar chart in our emotion evaluation system. The emotion radar chart can show the intensity of each emotion. This emotion evaluation system can help hearing-impaired to learn how to express emotions in speech naturally, just like ordinary people.
|
|---|