Interacting with Annotated and Synchronized Music Corpora on the Dezrann Web Platform

Abstract

Open datasets with annotated corpora are crucial to foster research in music information retrieval (MIR) studies and to disseminate knowledge towards musicians and the general public. Many groups have published corpora, at times accompanied by specific visualization interfaces. However, achieving a cohesive access to such a diverse range of data poses a significant challenge. Dezrann is an open-source web platform to interact with corpora while sharing music and music analysis in the form of scores, images, audio files (waveforms), video files, and annotations. We present how we render through this platform ten curated corpora published in the last years in the MIR community, gathering 1500+ pieces and 35000+ annotations. These corpora include works by Bach, Mozart, Beethoven, and Schubert, lieder from the 19th century by female composers (OpenScore Lieder), jazz solo transcriptions (Weimar Jazz Database), piano rolls (SUPRA), Georgian sacred songs (Erkomaishvili dataset), and Slovenian folk song ballads. Showing these corpora with the cross-modal synchronization enabled by the platform improves the way to hear, study, and annotate music. This opens up new possibilities for corpus annotation and analysis in musicology and computer music research, enhances music education, fosters the promotion of music diversity, and facilitates collaborative score editing and correction.