An Efficient Classification Algorithm for Music Mood Detection in Western and Hindi Music Using Audio Feature Extraction

Over the past decade, a lot of research has been done in audio content analysis for extracting various kinds of information, especially the moods it denotes, from an audio signal, because music expresses emotions in a concise and succinct way, yet in an effective way. People select music in congruence to their moods and emotions, making the need to classify music in accordance to moods more of a demand. Since different individuals have different perceptions about classifying music according to mood, it becomes a much more difficult task. This paper proposes an automated and efficient method to perceive the mood of any given music piece, or the "emotions" related to it, by drawing out a link between the spectral and harmonic features and human perception of music and moods. Features such as rhythm, harmony, spectral feature, and so on, are studied in order to classify the songs according to its mood, based on Thayer's model. The values of the quantified features are then compared against the threshold value using neural networks before classifying them according to different mood labels. The method analyzes many different features of the music piece, including spectra of beat and roughness, before classifying it under any mood. A total of 8 different moods are considered. In particular, the paper classifies both western and Indian Hindi film music, taking into consideration, a database of over 100 songs in total. The efficiency of this method was found to reach 94.44% at the best.