Several sound classifiers have been developed throughout the years. The accuracy provided by these classifiers is influenced by the features they use and the classification method implemented. While there are many approaches in sound feature extraction and in sound classification, most have been used to classify sounds with very different characteristics. Here, we propose a similar sound classifier that is able to distinguish sounds with very similar properties, namely sounds produced by objects with similar geometry and that only differ in material. The classifier applies independent component analysis to learn temporal and spectral features of the sounds, which are then used by a 1-nearest neighbor algorithm. We concluded that the features extracted in this way are powerful enough for classifying similar sounds. Finally, a user study shows that the classifier achieves better performance than humans in the classification of the sounds used here.
|Title of host publication||Lecture Notes in Computer Science|
|Publication status||Published - 1 Jan 2010|
|Event||International Conference on Image and Signal Processing (ICISP) - |
Duration: 1 Jan 2010 → …
|Conference||International Conference on Image and Signal Processing (ICISP)|
|Period||1/01/10 → …|