dc.contributor.advisor | © 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. | |
dc.contributor.author | O'Hanlon, K | |
dc.contributor.author | Benetos, E | |
dc.contributor.author | Dixon, S | |
dc.contributor.author | IEEE International Workshop on Machine Learning for Signal Processing (MLSP) | |
dc.date.accessioned | 2021-10-01T13:29:34Z | |
dc.date.available | 2021-08-15 | |
dc.date.available | 2021-10-01T13:29:34Z | |
dc.date.issued | 2021-10-25 | |
dc.identifier.uri | https://qmro.qmul.ac.uk/xmlui/handle/123456789/74336 | |
dc.description.abstract | Deep Learning (DL) has recently been applied successfully to the task of Cover Song Identification (CSI). Meanwhile, neural networks that consider music signal data structure in their design have been developed. In this paper, we propose a Pitch Class Key-Invariant Network, PiCKINet, for CSI. Like some other CSI networks, PiCKINet inputs a Constant-Q Transform (CQT) pitch feature. Unlike other such networks, large multi-octave kernels produce a latent representation with pitch class dimensions that are maintained throughout PiCKINet by key-invariant convolutions. PiCKINet is seen to be more effective, and efficient, than other CQT-based networks. We also propose an extended variant, PiCKINet+, that employs a centre loss penalty, squeeze and excite units, and octave swapping data augmentation. PiCKINet+ shows an improvement of ~17% MAP relative to the well-known CQTNet when tested on a set of ~16K tracks. | en_US |
dc.format.extent | ? - ? (6) | |
dc.publisher | IEEE | en_US |
dc.title | Detecting cover songs with pitch class key-invariant networks | en_US |
dc.type | Conference Proceeding | en_US |
dc.rights.holder | © 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. | |
pubs.notes | Not known | en_US |
pubs.publication-status | Accepted | en_US |
pubs.publisher-url | https://2021.ieeemlsp.org/ | en_US |
dcterms.dateAccepted | 2021-08-15 | |
rioxxterms.funder | Default funder | en_US |
rioxxterms.identifier.project | Default project | en_US |
qmul.funder | Development of next generation music recognition algorithm for content monitoring::Innovate UK | en_US |