Published October 31, 2023 | Version v2
Dataset Open

MuVi-Sync Dataset

  • 1. ROR icon Singapore University of Technology and Design

Description

Welcome to the MuVi-Sync dataset! This collection provides a rich array of features for both music and video elements. Here's a breakdown of the directory structure:

Music Features

  • vevo_chord: Chord feature data
  • vevo_note_density: Note density feature data
  • vevo_loudness: Loudness feature data

Video Features

  • vevo_scene_offset: Scene offset feature data
  • vevo_emotion: Emotion feature data
    • 5c_l14p: 5 emotion categories (exciting, fearful, tense, sad, relaxing)
    • 6c_l14p: 6 emotion categories (exciting, fearful, tense, sad, relaxing, neutral)
  • vevo_semantic: Semantic feature
  • vevo_motion: Motion feature

Others

  • vevo_meta:
    • idlist.txt: List of features, titles, and YouTube IDs
  • vevo: Original video files (.mp4) (on request by email: kjysmu@gmail.com)
  • vevo_audio: Original audio files (.mp3) (on request by email: kjysmu@gmail.com)

 

For detailed information about our dataset, please refer to our paper,"Video2Music: Suitable Music Generation from Videos using an Affective Multimodal Transformer model"  [ Paper ] | [ GitHub ]

Files

Files (356.0 MB)

Name Size Download all
md5:d6aee981e6e51d4a9c9301fbf234be79
356.0 MB Download