Mapping between dynamic markings and performed loudness: A machine learning approach
- Submitting institution
-
Queen Mary University of London
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 2564
- Type
- D - Journal article
- DOI
-
10.1080/17459737.2016.1193237
- Title of journal
- Journal of Mathematics and Music
- Article number
- -
- First page
- 149
- Volume
- 10
- Issue
- 2
- ISSN
- 1745-9745
- Open access status
- Compliant
- Month of publication
- August
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
3
- Research group(s)
-
-
- Citation count
- 4
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- A large-scale analysis of practical expressions of dynamic markings composers notate in a score. Machine learning techniques (decision trees, SVMs, ANNs) are applied to (i) predict loudness in recordings, and (ii) predict dynamic markings, based on known marking-loudness associations. This was the first computational and systematic analysis of performance practice on this scale. The research revealed counter-intuitive findings, like performer's approach to other mazurkas do not help predict performance of a new mazurka, and produced a new dataset of loudness and beat information for 2000 recordings of Chopin mazurkas that are freely available to advance performance science.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -