O'Reilly logo

Music Emotion Recognition by Homer H. Chen, Yi-Hsuan Yang

Stay ahead with the world's most comprehensive technology and business learning platform.

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, tutorials, and more.

Start Free Trial

No credit card required

Saunder January 24, 2011 10:39 book
1
Introduction
One of the most appealing functions of music is that it can convey emotion and
modulate a listener’s mood [90]. It is generally believed that music cannot be com-
posed, performed, or listened to without affection involvement [159]. Music can
bring us to tears, console us when we are grieving, and drive us to love. Music infor-
mation behavior studies have also identified emotion as an important criterion used
by people in music searching and organization. In this chapter, we describe the im-
portance of music emotion recognition, with examples that illustrate the application
of emotion-based music retrieval. After briefly describing the system components of
a typical music emotion recognition system, we move on to discuss the issues one
needs to consider when building the music emotion recognition system.
1.1 Importance of Music Emotion Recognition
Music plays an important role in human life, even more so in the digital age. Never
before has such a large collection of music been created and accessed daily by people.
The popularity of the Internet and the use of compact audio formats with near
CD quality, such as MP3 (MPEG-1 Audio Layer 3), have greatly contributed to
the tremendous growth of digital music libraries [340]. The prevailing context in
which we encounter music is ubiquitous, including those in which the most routine
activities of life take place: waking up, eating, housekeeping, shopping, studying,
exercising, driving, and so forth [159]. Conventionally, the management of music
collections is based on catalog metadata, such as artist name, album name, and song
title. As the amount of content continues to explode, this conventional approach may
be no longer sufficient. The way that music information is organized and retrieved
has to evolve to meet the ever increasing demand for easy and effective information
access [146,149,188,193].
1
Saunder January 24, 2011 10:39 book
2 Music Emotion Recognition
Music, as a complex acoustic and temporal structure, is rich in content and
expressivity [33]. According to [159], when an individual engages with music as a
composer, performer, or listener, a very board range of mental processes is involved,
including representational and evaluative. The representational process includes the
perception of meter, rhythm, tonality, harmony, melody, form, and style, whereas
the evaluative process includes the perception of preference, aesthetic experience,
mood, and emotion. The term evaluative is used because such processes are typically
both valenced and subjective. Both the representational and the evaluative processes
of music listening can be leveraged to enhance music retrieval. In [146], Huron
specifically points out that since the preeminent functions of music are social and
psychological, the most useful characterization of music for facilitating large-scale
music access would be based on four types of information: style, emotion, genre,
and similarity.
According to a study of social tagging on Last.fm [6], a popular commercial music
website, emotion tag is the third most frequent type of tags (second to genre and
locale) assigned to music pieces by online users [178]. Even though emotion-based
music retrieval was a new idea at that time, a survey conducted in 2004 showed
that about 28.2% of the participants identified emotion as an important criterion
in music seeking and organization (see Table 1.1) [188, 193]. Since then, emotion-
based music retrieval has received increasing attention in both academia and the
industry [140,145,159,217, 229, 364].
Table 1.1 Responses of 427 Subjects to the Question “When You Search
for Music or Music Information, How Likely Are You to Use the Following
Search/Browse Options?”
Positive Positive
Search/Browse By Rate Search/Browse By Rate
Singer/performer 96.2% Popularity 31.0%
Title of work(s) 91.7% Mood/emotional state 28.2%
Some words of the lyrics 74.0% Time period 23.8%
Music style/genre 62.7% Occasions to use 23.6%
Recommendations 62.2% Instrument(s) 20.8%
Similar artist(s) 59.3% Place/event where heard 20.7%
Similar music 54.2% Storyline of music 17.9%
Associated usage (ad, etc.) 41.9% Tempo 14.2%
Singing/humming 34.8% Record label 11.7%
Theme (main subject) 33.4% Publisher 6.0%
Source: Data from J.H. Lee and J.S. Down i.e. Proc. Int. Conf. Music Informa-
tion Retrieval 2004: 441446 [193].

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, interactive tutorials, and more.

Start Free Trial

No credit card required