Learning Musical Representations for Music Performance Question Answering

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Diao, Xingjian, Zhang, Chunhui, Wu, Tingxuan, Cheng, Ming, Ouyang, Zhongyu, Wu, Weiyi, Gui, Jiang
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917917784801280
author Diao, Xingjian
Zhang, Chunhui
Wu, Tingxuan
Cheng, Ming
Ouyang, Zhongyu
Wu, Weiyi
Gui, Jiang
author_facet Diao, Xingjian
Zhang, Chunhui
Wu, Tingxuan
Cheng, Ming
Ouyang, Zhongyu
Wu, Weiyi
Gui, Jiang
contents Music performances are representative scenarios for audio-visual modeling. Unlike common scenarios with sparse audio, music performances continuously involve dense audio signals throughout. While existing multimodal learning methods on the audio-video QA demonstrate impressive capabilities in general scenarios, they are incapable of dealing with fundamental problems within the music performances: they underexplore the interaction between the multimodal signals in performance and fail to consider the distinctive characteristics of instruments and music. Therefore, existing methods tend to answer questions regarding musical performances inaccurately. To bridge the above research gaps, (i) given the intricate multimodal interconnectivity inherent to music data, our primary backbone is designed to incorporate multimodal interactions within the context of music; (ii) to enable the model to learn music characteristics, we annotate and release rhythmic and music sources in the current music datasets; (iii) for time-aware audio-visual modeling, we align the model's music predictions with the temporal dimension. Our experiments show state-of-the-art effects on the Music AVQA datasets. Our code is available at https://github.com/xid32/Amuse.
format Preprint
id arxiv_https___arxiv_org_abs_2502_06710
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning Musical Representations for Music Performance Question Answering
Diao, Xingjian
Zhang, Chunhui
Wu, Tingxuan
Cheng, Ming
Ouyang, Zhongyu
Wu, Weiyi
Gui, Jiang
Computer Vision and Pattern Recognition
Multimedia
Sound
Audio and Speech Processing
Music performances are representative scenarios for audio-visual modeling. Unlike common scenarios with sparse audio, music performances continuously involve dense audio signals throughout. While existing multimodal learning methods on the audio-video QA demonstrate impressive capabilities in general scenarios, they are incapable of dealing with fundamental problems within the music performances: they underexplore the interaction between the multimodal signals in performance and fail to consider the distinctive characteristics of instruments and music. Therefore, existing methods tend to answer questions regarding musical performances inaccurately. To bridge the above research gaps, (i) given the intricate multimodal interconnectivity inherent to music data, our primary backbone is designed to incorporate multimodal interactions within the context of music; (ii) to enable the model to learn music characteristics, we annotate and release rhythmic and music sources in the current music datasets; (iii) for time-aware audio-visual modeling, we align the model's music predictions with the temporal dimension. Our experiments show state-of-the-art effects on the Music AVQA datasets. Our code is available at https://github.com/xid32/Amuse.
title Learning Musical Representations for Music Performance Question Answering
topic Computer Vision and Pattern Recognition
Multimedia
Sound
Audio and Speech Processing
url https://arxiv.org/abs/2502.06710