New Directions in Digital Jazz Studies: Music Information Retrieval and AI Support for Jazz Scholarship in Digital Archives

点击翻译按钮获取中文摘要

Music research has developed a broad range of methods, where systematic musicology, ethnomusicology, or music psychology, have developed as "data oriented empirical research", which benefits from computing methods. In ethnomusicology particularly, there has been a recent growing interest in computational musicology and its application to audio data collections. Similarly, the empirical study of performance of Western music, such as timing, dynamics and timbre and their relation to musical structure has a long tradition. However, despite recent progress, this type of music research has so far been mostly limited to relatively small datasets, because of technological and legal limitations. Musicology and music history as a humanistic disciplines have so far had limited benefit from computational and dda methods, because a) the algorithms and technologies require too much technical expertise and b) the objects of study (recordings, scores, manuscripts and other artefacts), although often available in digitised form, cannot be accessed with the relevant tools, such as analysis algorithms, content and context based intelligent search, and user friendly systems and interfaces. This project will change this limitation and explore new directions in jazz research by developing novel methods and tools based on existing algorithms (developed by project partners in previous research) and applying these tools to jazz studies. Digital material in archives will be made more accessible and connected between different archives, audio recording will be analysed with state-of-the-art algorithms and contextualized with linked data, and research with digital tools will be made shareable. By combining these elements, novel methods can be developed and novel insights can be gained. We will collect requirements in participating archives and from the jazz research community to ensure that our work matches the needs and interests of digital humanities applied to jazz in practice. Based on the requirements, we will curate datasets, ensuring quality of the data and metadata, before pre-processing the datasets to extract symbolic information from audio recordings. Building on existing algorithms and tools, we will develop systems and interfaces that enable musically meaningful search, analysis and discovery, that goes beyond standard catalog search. We have several well-known jazz scholars in the project, who will use the datasets and tools we develop to develop case studies in humanistic research of the 'long tail' of lesser known recordings present in archives. These case studies will exemplify new directions in jazz studies that combine data-driven 'distant reading' with human expertise in 'close reading' with the help of the tool developed in this project . Musical questions addressed in this work will include how personal styles in relation to specific other players as well as to general trends in mainstream jazz in the US. It will also include the study of local jazz traditions, specifically in Scotland, and comparative studies overarching the two archives and exemplifying tools and methods for connecting digital archives that are on different continents.The results of these analyses will be made available in the form of highly interactive 'rich research workflows' (RRW). The rich research log will contain research queries, result sets, contextual information, visualisation and annotations. The RRW can be a medium for sharing working knowledge about research in digital archives similar to the increasingly popular sharing of code in computing research. All results will be made available as open data/open source software. We feel that this project has the potential to bring together communities from musicology, libraries and archives, and computing to mutual benefit.