In this research program, we have created a database which includes folk music from Cyprus, Turkey and middle east, Western melodies and Byzantine music. In this database, we include recordings that were made by the researchers of our group using professional audio equipment. We have created computational tools for analysing and extracting low level, mid level and high level features that are used for similarity detection between the songs in our database. Examples of high level features that are created include note segmentation, ornamentation detection, tonic and scale identification. We have used techniques for tonal similarity between pairs of songs using pitch histograms.
Additionally. we studied the timbre characteristics of a collection of songs from our database and we followed classification and clustering approaches to model and compare groups of songs. A number of matlab scripts are provided in a user friendly environment for visualization of song characteristics, pair wise comparison, note segmentation and other attributes. This research was funded from the Republic of Cyprus through the Cyprus research promotion foundation by the research grant ΑΝΘΡΩΠΙΣΤΙΚΕΣ / ΑΝΘΡΩ / 0311(BE) / 19, and also supported by the University of Cyprus.
© 2017-2019 Department of Computer Science - University of Cyprus
Created by ITS - Based on Minimal by W3layouts