Music Genre Classification Base On Clustered Melody Patterns
碩士 === 國立臺灣科技大學 === 資訊管理系 === 100 === This paper proposes a scheme of using melodic patterns to classify the musical genres. Three types of statistical classification approaches are compared, correlation-based classifier, artificial neural network(ANN), K-nearest neighbor classifier. In the baseline...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2012
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/rg47e4 |
| Summary: | 碩士 === 國立臺灣科技大學 === 資訊管理系 === 100 === This paper proposes a scheme of using melodic patterns to classify the musical genres. Three types of statistical classification approaches are compared, correlation-based classifier, artificial neural network(ANN), K-nearest neighbor classifier. In the baseline experiment, ANN can achieve the highest accuracy of 67.5%, while the correlation-based classifier proposed in this paper the accuracy of 66.2%.
The correlation-based classifier were there improved by smoothing the statistical of correlations based on clustered melodic patterns. The accuracy of 70.0% can be achieved for correlation-based classifier after applying the smoothing approach. Finally, a scheme of conditional smoothing by considering the amount of training data can be further used to improve the accuracy up to 70.67%.
|
|---|