Developing a personalised Music Affect Recommender System.

点击翻译按钮获取中文摘要

The project aims to develop a personalised music recommender system using perceived tone quality, affect and liking. Recommender systems using prior verbal annotations and ratings are common (Amazon) but inappropriate for less popular music by unfamiliar artists, which lacks social use data. The project intends to build on work into perception of musical affect and its relation to loudness and tone quality; and the automation of the organisation of digital libraries both by labels and acoustic content. Developing this, the project seeks to create a model that gives recommendations which accounts for an individual's preferences based on acoustic content, affect and liking. The system will be designed to update rapidly and to encourage exploration of familiar and unfamiliar music.