{"id":{"repo_id":"cambridge","oai_identifier":"oai:www.repository.cam.ac.uk:1810/397920"},"canonical_url":"https://search.dev.ndltd.org/etd/cambridge/oai:www.repository.cam.ac.uk:1810/397920","repository":{"repo_id":"cambridge","name":"Cambridge University","base_url":"https://api.repository.cam.ac.uk/server/oai/request"},"display":{"title":"Computational Modelling of Jazz Improvisation","abstract":"Jazz is a highly improvised form of Western music that so far has received limited scientific attention. This is due in part both to the difficulty of analysing music for which written scores are generally not available and of crafting statistical models that are likely to be meaningful for those who engage with this music. This thesis reports a research programme of five studies that address these problems by first developing a large dataset of annotated jazz recordings and then analysing this and other similar datasets using a variety of computational methods. First, we describe the construction of the Jazz Trio Database (JTD). This is an open source dataset of 45 hours of jazz rhythm section performances with symbolic annotations, comprising both quarter-note downbeats and beats, alongside MIDI for the pianist. Second, we construct a series of supervised-learning models that learn to identify particular JTD pianists from features relating to their use of harmony, melody, rhythm, and dynamics. We then use the decision functions learned by these models to disentangle the different elements that contribute towards defining musical style in jazz. Third, we refine the scope of the previous study to focus on modelling the relationship between rhythm and musical style. We extract a range of rhythmic features (including those relating to “feel”, “swing”, and “interaction”) from JTD and use these to objectively test accounts of jazz rhythmic style given in prior musicological and ethnographic writing. Fourth, we model the mechanisms by which ensemble jazz improvisations actually take place. We use several features from the previous studies to explore the strategies that five duos of professional musicians employ to coordinate with one another under experimental conditions that are designed to disrupt the tight temporal coordination of action typically involved in group jazz performances. Finally, we introduce a generative model that produces music in the styles of particular performers and jazz subgenres, which we train using data and techniques introduced across the entire thesis. Our studies are accompanied by open-source implementations of our models, data, and research software, alongside numerous interactive web applications. We hope these will facilitate future computational research into musical improvisation — in jazz, and beyond.","abstract_html":"Jazz is a highly improvised form of Western music that so far has received limited scientific attention. This is due in part both to the difficulty of analysing music for which written scores are generally not available and of crafting statistical models that are likely to be meaningful for those who engage with this music. This thesis reports a research programme of five studies that address these problems by first developing a large dataset of annotated jazz recordings and then analysing this and other similar datasets using a variety of computational methods. First, we describe the construction of the Jazz Trio Database (JTD). This is an open source dataset of 45 hours of jazz rhythm section performances with symbolic annotations, comprising both quarter-note downbeats and beats, alongside MIDI for the pianist. Second, we construct a series of supervised-learning models that learn to identify particular JTD pianists from features relating to their use of harmony, melody, rhythm, and dynamics. We then use the decision functions learned by these models to disentangle the different elements that contribute towards defining musical style in jazz. Third, we refine the scope of the previous study to focus on modelling the relationship between rhythm and musical style. We extract a range of rhythmic features (including those relating to “feel”, “swing”, and “interaction”) from JTD and use these to objectively test accounts of jazz rhythmic style given in prior musicological and ethnographic writing. Fourth, we model the mechanisms by which ensemble jazz improvisations actually take place. We use several features from the previous studies to explore the strategies that five duos of professional musicians employ to coordinate with one another under experimental conditions that are designed to disrupt the tight temporal coordination of action typically involved in group jazz performances. Finally, we introduce a generative model that produces music in the styles of particular performers and jazz subgenres, which we train using data and techniques introduced across the entire thesis. Our studies are accompanied by open-source implementations of our models, data, and research software, alongside numerous interactive web applications. We hope these will facilitate future computational research into musical improvisation — in jazz, and beyond.","abstract_has_math":false,"creators":["Cheston, Huw"],"institution":"University of Cambridge","degree_name":"Doctor of Philosophy (PhD)","degree_level":"Doctoral","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Harrison, Peter"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-06-20","date_published":"2025-06-20","updated_at":"2026-07-24T01:33:11Z","subjects":["music information retrieval","machine learning","jazz improvisation","explainable artificial intelligence","corpus analysis"],"languages":["eng"],"rights":[],"rights_urls":["https://www.repository.cam.ac.uk/bitstreams/97a81f07-f8cf-4b95-b52b-d3c28772e693/download","http://purl.org/NET/rdflicense/allrightsreserved"],"identifier_entries":[]},"links":{"outbound_url":"https://doi.org/10.17863/CAM.126908","outbound_label":"DOI","outbound_source":"dc:identifier.doi"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Harrison, Peter"]},{"key":"dc:contributor.sponsor","label":"Sponsor","values":["This work was supported by a Vice-Chancellor’s Award from the Cambridge Trust. Additional funding for the research described in Chapter 5 was received from a Project Incubation Award by Cambridge Digital Humanities."]},{"key":"dc:creator","label":"Author","values":["Cheston, Huw"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2025-06-20"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["University of Cambridge"]},{"key":"dc:relation.isreferencedby.uri","label":"Dc Relation Isreferencedby URI","values":["https://www.repository.cam.ac.uk/handle/1810/397920"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["Doctoral"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["Doctor of Philosophy (PhD)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["music information retrieval","machine learning","jazz improvisation","explainable artificial intelligence","corpus analysis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["https://www.repository.cam.ac.uk/bitstreams/97a81f07-f8cf-4b95-b52b-d3c28772e693/download","http://purl.org/NET/rdflicense/allrightsreserved"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.17863/CAM.126908"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://www.repository.cam.ac.uk/bitstreams/862e5b1b-a610-4834-98a2-2b2fc3019c04/download"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Jazz is a highly improvised form of Western music that so far has received limited scientific attention. This is due in part both to the difficulty of analysing music for which written scores are generally not available and of crafting statistical models that are likely to be meaningful for those who engage with this music. This thesis reports a research programme of five studies that address these problems by first developing a large dataset of annotated jazz recordings and then analysing this and other similar datasets using a variety of computational methods. First, we describe the construction of the Jazz Trio Database (JTD). This is an open source dataset of 45 hours of jazz rhythm section performances with symbolic annotations, comprising both quarter-note downbeats and beats, alongside MIDI for the pianist. Second, we construct a series of supervised-learning models that learn to identify particular JTD pianists from features relating to their use of harmony, melody, rhythm, and dynamics. We then use the decision functions learned by these models to disentangle the different elements that contribute towards defining musical style in jazz. Third, we refine the scope of the previous study to focus on modelling the relationship between rhythm and musical style. We extract a range of rhythmic features (including those relating to “feel”, “swing”, and “interaction”) from JTD and use these to objectively test accounts of jazz rhythmic style given in prior musicological and ethnographic writing. Fourth, we model the mechanisms by which ensemble jazz improvisations actually take place. We use several features from the previous studies to explore the strategies that five duos of professional musicians employ to coordinate with one another under experimental conditions that are designed to disrupt the tight temporal coordination of action typically involved in group jazz performances. Finally, we introduce a generative model that produces music in the styles of particular performers and jazz subgenres, which we train using data and techniques introduced across the entire thesis. Our studies are accompanied by open-source implementations of our models, data, and research software, alongside numerous interactive web applications. We hope these will facilitate future computational research into musical improvisation — in jazz, and beyond."]},{"key":"dc:format.checksum.md5","label":"Dc Format Checksum Md5","values":["4ba8d387ffc60b2d24ef722a317462b3","87eda9de84448d1f82354d60eee3eb5f"]},{"key":"dc:title","label":"Title","values":["Computational Modelling of Jazz Improvisation"]}]}],"canonical_facts":{"dc:contributor.advisor":["Harrison, Peter"],"dc:contributor.sponsor":["This work was supported by a Vice-Chancellor’s Award from the Cambridge Trust. Additional funding for the research described in Chapter 5 was received from a Project Incubation Award by Cambridge Digital Humanities."],"dc:creator":["Cheston, Huw"],"dc:date.issued":["2025-06-20"],"dc:description.abstract":["Jazz is a highly improvised form of Western music that so far has received limited scientific attention. This is due in part both to the difficulty of analysing music for which written scores are generally not available and of crafting statistical models that are likely to be meaningful for those who engage with this music. This thesis reports a research programme of five studies that address these problems by first developing a large dataset of annotated jazz recordings and then analysing this and other similar datasets using a variety of computational methods. First, we describe the construction of the Jazz Trio Database (JTD). This is an open source dataset of 45 hours of jazz rhythm section performances with symbolic annotations, comprising both quarter-note downbeats and beats, alongside MIDI for the pianist. Second, we construct a series of supervised-learning models that learn to identify particular JTD pianists from features relating to their use of harmony, melody, rhythm, and dynamics. We then use the decision functions learned by these models to disentangle the different elements that contribute towards defining musical style in jazz. Third, we refine the scope of the previous study to focus on modelling the relationship between rhythm and musical style. We extract a range of rhythmic features (including those relating to “feel”, “swing”, and “interaction”) from JTD and use these to objectively test accounts of jazz rhythmic style given in prior musicological and ethnographic writing. Fourth, we model the mechanisms by which ensemble jazz improvisations actually take place. We use several features from the previous studies to explore the strategies that five duos of professional musicians employ to coordinate with one another under experimental conditions that are designed to disrupt the tight temporal coordination of action typically involved in group jazz performances. Finally, we introduce a generative model that produces music in the styles of particular performers and jazz subgenres, which we train using data and techniques introduced across the entire thesis. Our studies are accompanied by open-source implementations of our models, data, and research software, alongside numerous interactive web applications. We hope these will facilitate future computational research into musical improvisation — in jazz, and beyond."],"dc:format.checksum.md5":["4ba8d387ffc60b2d24ef722a317462b3","87eda9de84448d1f82354d60eee3eb5f"],"dc:identifier.doi":["https://doi.org/10.17863/CAM.126908"],"dc:identifier.uri":["https://www.repository.cam.ac.uk/bitstreams/862e5b1b-a610-4834-98a2-2b2fc3019c04/download"],"dc:language":["eng"],"dc:publisher.institution":["University of Cambridge"],"dc:relation.isreferencedby.uri":["https://www.repository.cam.ac.uk/handle/1810/397920"],"dc:rights":["https://www.repository.cam.ac.uk/bitstreams/97a81f07-f8cf-4b95-b52b-d3c28772e693/download","http://purl.org/NET/rdflicense/allrightsreserved"],"dc:subject":["music information retrieval","machine learning","jazz improvisation","explainable artificial intelligence","corpus analysis"],"dc:title":["Computational Modelling of Jazz Improvisation"],"dc:type":["Thesis"],"dc:type.qualificationlevel":["Doctoral"],"dc:type.qualificationname":["Doctor of Philosophy (PhD)"]},"updated_at":"2026-07-24T01:33:11Z"}