This presentation reviews different representations and algorithms dedicated to the processing of symbolic musical data, enabling the highlighting of different aspects of musical language that appears within musical scores. The notion of musical language is approached through essential components of scores, including structure, texture, or annotations relative to instrumental practice in the specific case of guitar tablatures. We will then discuss the adaptation of machine learning algorithms designed in the field of Natural Language Processing (NLP) to process musical scores, and how this approach questions us regarding the association of music with a particular kind of language. We will finally present an experiment of musical human/computer co-composition involving these concepts, which has been submitted to the 2020 AI Song Contest.
Associate Professor Louis Bigo
University of Lille
Prof. Louis Bigo is an associate professor in computer science at University of Lille (France) since 2016. He is conducting research in the field of music informatics within the Algomus team at CRIStAL laboratory. His research focusses on the elaboration of mathematical and machine learning models intended to assist the analysis and composition of musical scores. He defended his PhD in 2013 on the topic “Symbolic musical representations and spatial computing” (IRCAM, LACL, Université Paris-Est, France) and did a post-doctorate from 2014 to 2016 in machine learning and music generation (University of the Basque country, San-Sebastian, Spain).