MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training

Fuente: arXiv
Saved in:
Bibliographic Details
Main 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
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910762685956096
author 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
author_facet 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
contents Self-supervised learning (SSL) has recently emerged as a promising paradigm for training generalisable models on large-scale data in the fields of vision, text, and speech. Although SSL has been proven effective in speech and audio, its application to music audio has yet to be thoroughly explored. This is partially due to the distinctive challenges associated with modelling musical knowledge, particularly tonal and pitched characteristics of music. To address this research gap, we propose an acoustic Music undERstanding model with large-scale self-supervised Training (MERT), which incorporates teacher models to provide pseudo labels in the masked language modelling (MLM) style acoustic pre-training. In our exploration, we identified an effective combination of teacher models, which outperforms conventional speech and audio approaches in terms of performance. 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). Furthermore, we explore a wide range of settings to overcome the instability in acoustic language model pre-training, which allows our designed paradigm to scale from 95M to 330M parameters. 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.
format Preprint
id arxiv_https___arxiv_org_abs_2306_00107
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training
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
Sound
Artificial Intelligence
Computation and Language
Machine Learning
Audio and Speech Processing
Self-supervised learning (SSL) has recently emerged as a promising paradigm for training generalisable models on large-scale data in the fields of vision, text, and speech. Although SSL has been proven effective in speech and audio, its application to music audio has yet to be thoroughly explored. This is partially due to the distinctive challenges associated with modelling musical knowledge, particularly tonal and pitched characteristics of music. To address this research gap, we propose an acoustic Music undERstanding model with large-scale self-supervised Training (MERT), which incorporates teacher models to provide pseudo labels in the masked language modelling (MLM) style acoustic pre-training. In our exploration, we identified an effective combination of teacher models, which outperforms conventional speech and audio approaches in terms of performance. 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). Furthermore, we explore a wide range of settings to overcome the instability in acoustic language model pre-training, which allows our designed paradigm to scale from 95M to 330M parameters. 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.
title MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training
topic Sound
Artificial Intelligence
Computation and Language
Machine Learning
Audio and Speech Processing
url https://arxiv.org/abs/2306.00107