Evaluation of the Audio Beat Tracking System BeatRoot

Bibliographic Data

ID5304983
AuthorsSimon Dixon (0000-0002-6098-481X, Queen Mary University of London, corresponding author)
Year2007
Volume36
Issue1
Pages39-50
Publication date2007-03-01
Peer ReviewedYes
Open AccessNo
TypeARTICLE
VenueJournal of New Music Research (JOURNAL)
Journal identifiersISSN: 0929-8215 • E-ISSN: 1744-5027
PublisherInforma UK Limited (PUBLISHER • GB)
DOI10.1080/09298210701653310
OpenAlexW2134679199
LanguageEN
Citations received10
References cited15

BeatRoot is an interactive beat tracking and metrical annotation system which has been used for several years in studies of performance timing. This paper describes improvements to the original system and a large-scale evaluation and analysis of the system's performance. The beat tracking algorithm remains largely unchanged: BeatRoot uses a multiple agent architecture which simultaneously considers several different hypotheses concerning the rate and placement of musical beats, resulting in accurate tracking of the beat, quick recovery from errors, and graceful degradation in cases where the beat is only weakly implied by the data. The original time domain onset detection algorithm, which caused problems for beat tracking music without prominent drums, has been replaced with a more robust onset detector based on spectral flux. Other new features include annotation of multiple metrical levels and phrase boundaries, as well as improvements in the user interface. The software has been rewritten entirely in Java, and it runs on all major operating systems. BeatRoot is evaluated on a new test set of 1360 musical excerpts from a wide range of Western musical styles, and the results are compared with other evaluations such as the MIREX 2006 Audio Beat Tracking Evaluation

Annotation · Speech recognition · Computer Science · Music and Audio Processing · Music Technology and Sound Studies · Speech and Audio Processing · Artificial Intelligence

Tools

Open DOISci-Hub