Swiftlet sound identification using vector quantization and minimum distance classifier
There are high demand on swiftlet nest as it benefits in health, cosmetic and food industry. Therefore, the study about technologies and method to increase their production in swiftlet farming using sound technology is needed. In the real situation, the classification of swiftlet sound is evaluated...
| Main Authors: | , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English English |
| Published: |
2016
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/18733/ http://umpir.ump.edu.my/id/eprint/18733/1/Swiftlet%20Sound%20Identification%20using%20Vector%20Quantization%20and%20Minimum%20Distance%20Classifier.pdf http://umpir.ump.edu.my/id/eprint/18733/2/Swiftlet%20Sound%20Identification%20using%20Vector%20Quantization%20and%20Minimum%20Distance%20Classifier%201.pdf |
| Summary: | There are high demand on swiftlet nest as it benefits in health, cosmetic and food industry. Therefore, the study about technologies and method to increase their production in swiftlet farming using sound technology is needed. In the real situation, the classification of swiftlet sound is evaluated by human expert using try and error method at swiftlet house. However, this required high level of human skill and prone to mistake. In this work, we present an automatic swiftlet sound identification using vector quantization and minimum distance classifier. Firstly, swiftlet sound extracted using mel-frequency cepstral coefficient. Secondly, vector quantization with codebook size is 8,16,32 and 64 and minimum distance classifier was used for the sound classification. Finally, performance of the system was measured by in three type swiftlets, baby, adults and colony type. It shows that, the highest identification was ?? when using what and what linear predictive cepstral coefficient features change to mel frequency cepstral coefficient additional deltaacceleration features. |
|---|