Show simple item record

Early MFCC And HPCP Fusion for Robust Cover Song Identification

dc.contributor.author Tralie, Christopher
dc.date.accessioned 2017-12-11T15:06:08Z
dc.date.available 2017-12-11T15:06:08Z
dc.date.issued 2017-12-11
dc.identifier http://arxiv.org/abs/1707.04680v1
dc.identifier.uri http://hdl.handle.net/10161/15840
dc.description.abstract While most schemes for automatic cover song identification have focused on note-based features such as HPCP and chord profiles, a few recent papers surprisingly showed that local self-similarities of MFCC-based features also have classification power for this task. Since MFCC and HPCP capture complementary information, we design an unsupervised algorithm that combines normalized, beat-synchronous blocks of these features using cross-similarity fusion before attempting to locally align a pair of songs. As an added bonus, our scheme naturally incorporates structural information in each song to fill in alignment gaps where both feature sets fail. We show a striking jump in performance over MFCC and HPCP alone, achieving a state of the art mean reciprocal rank of 0.87 on the Covers80 dataset. We also introduce a new medium-sized hand designed benchmark dataset called "Covers 1000," which consists of 395 cliques of cover songs for a total of 1000 songs, and we show that our algorithm achieves an MRR of 0.9 on this dataset for the first correctly identified song in a clique. We provide the precomputed HPCP and MFCC features, as well as beat intervals, for all songs in the Covers 1000 dataset for use in further research.
dc.format.extent 11 pages, 7 figures
dc.subject cs.IR
dc.subject cs.IR
dc.subject cs.SD
dc.subject H.5.5
dc.title Early MFCC And HPCP Fusion for Robust Cover Song Identification
dc.type Journal article
pubs.author-url http://arxiv.org/abs/1707.04680v1
pubs.notes Proceedings of The International Society for Music Information Retrieval (ISMIR) 2017
pubs.organisational-group Duke
pubs.organisational-group Mathematics
pubs.organisational-group Staff
pubs.organisational-group Temp group - logins allowed
pubs.organisational-group Trinity College of Arts & Sciences


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record