Affective analysis of musical chords

Music invokes emotions in humans and hence sentiment extraction in music has been researched for a long time. This paper focuses on 2 research goals (RGs): RG1: Identifying and analyzing emotions associated with different musical chords. RG2: Suggesting a technique to compute Evaluation, Potency and Activity (EPA) [31] values for musical chords. For RG1 a user study is conducted wherein 30 people are asked to name a song under two emotional categories - “Happiness” and “Sadness”. Chord progression of each song is determined using the Chordify Web service and the frequency of occurrence of the chords under the two emotional categories is calculated and the trends are analyzed. For RG2, EPA values for chords are computed by utilizing the results of RG1 to calculate the probability of chord, given an emotion Pr(Chord|Emotion). This data is fed into the proposed formula to determine EPA values associated with different chords. Thereafter, application of these results to existing sentiment extraction models is suggested.