EMOPIA (A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation)

EMOPIA（发音为 'yee-mò-pi-uh'）数据集是一个共享的多模式（音频和 MIDI）数据库，专注于流行钢琴音乐中的情感感知，以促进与音乐情感相关的各种任务的研究。该数据集包含来自 387 首歌曲的 1,087 个音乐片段和由四个专用注释器注释的片段级情感标签。

EMOPIA (pronounced /ˈjiːmoʊpiə/) is a shared multimodal (audio and MIDI) database focusing on emotion perception in popular piano music, aimed at facilitating research on various tasks related to musical emotion. This dataset contains 1,087 musical excerpts from 387 songs, along with excerpt-level emotion labels annotated by four dedicated annotators.

搜集汇总

背景与挑战

背景概述

EMOPIA是一个多模态流行钢琴数据集，专注于情感识别和情感驱动的音乐生成研究。该数据集包含来自387首歌曲的1,087个音乐片段，每个片段都带有由专业注释器标注的情感标签。