Approximate Searching for Music Data in Real-Valued Feature Indexing1

Abstract

For the music database retrieval, the researchers extract the features, such as melodies, rhythms and chords, from the music data and develop indices to support the quick music search. Several reports have pointed out that these features can be transformed and represented in the forms of music feature strings or numeric values such that string indexing or numeric indexing is created, respectively, for music retrieval. Recently, A real-valued feature indexing technique has been proposed for music data retrieval. This approach has demonstrated that it can support music searching more efficiently than existing schemes. However, there is only exactly matching for music retrieval discussed. The approximate searching for music database has not received many attentions yet. In this paper, we will propose an approximate searching scheme for music data retrieval in a real-valued feature index. The experimental results show that our approach outperforms existing scheme for music approximate searching.

Categorization of Alpha Wave Music by SOM Reduction

“…The contents of music provided diverse musical features such as: melody, rhythm and chords, etc., which may represent a music genre. Therefore, content-based music retrieval and classification are important research fields for music databases [8,9,10,11]. In this research, we will investigate the features of the alpha wave music for analyzing and tries to categorize the alpha wave music by Self Organizing Map (SOM) approach [12,13,14].…”