CITATION — REFERENCE ENTRY
MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training — arXiv
- Key
- li2023-mert
- Authors
- Li, Yizhi; Yuan, Ruibin; Zhang, Ge; Ma, Yinghao; Chen, Xingran; Yin, Hanzhi; Xiao, Chenghao; Lin, Chenghua; Ragni, Anton; Benetos, Emmanouil; Gyenge, Norbert; Dannenberg, Roger; Liu, Ruibo; Chen, Wenhu; Xia, Gus; Shi, Yemin; Huang, Wenhao; Wang, Zili; Guo, Yike; Fu, Jie
- Issued
- 2023-5-31
- Type
- article
- Container
- arXiv
Raw CSL JSON
{
"URL": "https://arxiv.org/abs/2306.00107",
"type": "article",
"title": "MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training",
"author": [
{
"given": "Yizhi",
"family": "Li"
},
{
"given": "Ruibin",
"family": "Yuan"
},
{
"given": "Ge",
"family": "Zhang"
},
{
"given": "Yinghao",
"family": "Ma"
},
{
"given": "Xingran",
"family": "Chen"
},
{
"given": "Hanzhi",
"family": "Yin"
},
{
"given": "Chenghao",
"family": "Xiao"
},
{
"given": "Chenghua",
"family": "Lin"
},
{
"given": "Anton",
"family": "Ragni"
},
{
"given": "Emmanouil",
"family": "Benetos"
},
{
"given": "Norbert",
"family": "Gyenge"
},
{
"given": "Roger",
"family": "Dannenberg"
},
{
"given": "Ruibo",
"family": "Liu"
},
{
"given": "Wenhu",
"family": "Chen"
},
{
"given": "Gus",
"family": "Xia"
},
{
"given": "Yemin",
"family": "Shi"
},
{
"given": "Wenhao",
"family": "Huang"
},
{
"given": "Zili",
"family": "Wang"
},
{
"given": "Yike",
"family": "Guo"
},
{
"given": "Jie",
"family": "Fu"
}
],
"issued": {
"date-parts": [
[
2023,
5,
31
]
]
},
"number": "2306.00107",
"container-title": "arXiv"
}
Claims
-
MERT uses masked-language-modelling-style acoustic pretraining with pseudo-labels from an RVQ-VAE acoustic teacher and a CQT-based musical teacher, and was evaluated on 14 music understanding tasks.
"This combination includes an acoustic teacher based on Residual Vector Quantisation - Variational AutoEncoder (RVQ-VAE) and a musical teacher based on the Constant-Q Transform (CQT). [...] Experimental results indicate that our model can generalise and perform well on 14 music understanding tasks and attain state-of-the-art (SOTA) overall scores."
Available in