Emotion and themes recognition in music utilising convolutional and recurrent neural networks



Amiriparian, S, Gerczuk, M, Coutinho, E ORCID: 0000-0001-5234-1497, Baird, A, Ottl, S, Milling, M and Schuller, B
(2019) Emotion and themes recognition in music utilising convolutional and recurrent neural networks. .

[img] Text
MediaEval_2019_Music_Emotion_Recognition.pdf - Accepted Version

Download (500kB) | Preview

Abstract

Emotion is an inherent aspect of music, and associations to music can be made via both life experience and specific musical techniques applied by the composer. Computational approaches for music recognition have been well-established in the research community; however, deep approaches have been limited and not yet comparable to conventional approaches. In this study, we present our fusion system of end-to-end convolutional recurrent neural networks (CRNN) and pre-trained convolutional feature extractors for music emotion and theme recognition1. We train 9 models and conduct various late fusion experiments. Our best performing model (team name: AugLi) achieves 74.2 % ROC-AUC on the test partition which is 1.6 percentage points over the baseline system of the MediaEval 2019 Emotion & Themes in Music task.

Item Type: Conference or Workshop Item (Unspecified)
Depositing User: Symplectic Admin
Date Deposited: 01 Oct 2019 07:21
Last Modified: 23 Jun 2021 06:10
URI: https://livrepository.liverpool.ac.uk/id/eprint/3056460