Lyrics Mining for Music Meta-Data Estimation - Artificial Intelligence Applications and Innovations Access content directly
Conference Papers Year : 2016

Lyrics Mining for Music Meta-Data Estimation


Music meta-data comprise a number of structured attributes that provide descriptive annotations such as singer, author, genre and date of a song deposited in a digital library. While they provide a crucial knowledge to represent music entry in current information retrieval and recommendation systems applications, they suffer from two limitations in practice. First, they may contain missing or wrong attributes due to incomplete submissions. Second, available attributes may not suffice to characterize the music entry for the objective of the retrieval or recommendation task being considered. Here, we offer an automated way of estimating the meta-data of a song using its lyrics content. We focus on attributing the author, genre and release date of songs solely based on the lyrics information. To this end, we introduce a complete text classification framework which takes raw lyrics data as input and report estimated meta-data attributes. The performance of the system is evaluated based on its retrieval ability on a large dataset of Turkish songs, which was gathered in this study and made publicly available. The results promote the use of such technique as a complementary tool in organizing music repositories and implementing music information retrieval systems.
Fichier principal
Vignette du fichier
430537_1_En_47_Chapter.pdf (280.23 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-01557589 , version 1 (06-07-2017)





Hasan Oğul, Başar Kirmaci. Lyrics Mining for Music Meta-Data Estimation. 12th IFIP International Conference on Artificial Intelligence Applications and Innovations (AIAI), Sep 2016, Thessaloniki, Greece. pp.528-539, ⟨10.1007/978-3-319-44944-9_47⟩. ⟨hal-01557589⟩
140 View
169 Download



Gmail Facebook X LinkedIn More