Name
Abstract | ||
|---|---|---|
In order to improve the automated retrieval of similar songs, we need to develop an estimation method which can measure their subjective similarity. In this study, we assume that subjective similarity of songs is determined by both the acoustical similarity of the songs and the individuality of the listener. We focus mainly on the individuality of listeners, and use knowledge about this individuality to develop a subjective similarity estimation model.The results of our previous study suggest that the likelihood of someone judging two songs (a musical pair) as "similar" is influenced by the individual characteristics (individuality) of the listener. In this paper we refer to the likelihood of judging songs to be similar as the "tolerance" of the listener, and propose a model of subjective similarity evaluation which takes individual tolerance into account. In our experiment, we estimate listeners' tolerance using subjective musical similarity evaluation data. We also conduct an experiment using a much smaller amount of similarity evaluation data to estimate tolerance, as this would be desirable for practical applications. |
Year | Venue | Keywords |
|---|---|---|
2013 | 2013 PROCEEDINGS OF THE 21ST EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO) | music similarity, subjective similarity, similarity evaluation, individuality |
Field | DocType | Citations |
Computer science,Musical,Speech recognition | Conference | 0 |
PageRank | References | Authors |
0.34 | 4 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Shota Kawabuchi | 1 | 0 | 0.34 |
| Chiyomi Miyajima | 2 | 345 | 45.71 |
| Norihide Kitaoka | 3 | 277 | 43.70 |
| Kazuya Takeda | 4 | 1301 | 195.60 |