Mô Hình GMM Định Danh Tự Động Một Số Làn Điệu Dân Ca Quan Họ Bắc Ninh

Abstract

Automatic identification of music genre is of great importance in automating the process of storing, organizing and searching for vast amounts of information about music. Vietnam has folk songs that are very rich for all three regions of North -Central -South, including Quan ho Bac Ninh folk songs. This paper presents a method for identifying some of Quan ho Bac Ninh folk songs using the Gaussian Mixture Model (GMM) with model parameters including Mel Frequency Cepstral Coefficients (MFCCs), energy and fundamental frequency F0. The experiment results showed that the exact identification scores depend on the number of mixture components. The use of fundamental frequency information increased considerably the exact indentification score.