Abstract
This paper presents a model capable of learning the rhythmic characteristics of a music signal through unsupervised learning. The model learns a multi-layer hierarchy of rhythmic patterns ranging from simple structures on lower layers to more complex patterns on higher layers. The learned hierarchy is fully transparent, which enables observation and explanation of the structure of the learned patterns. The model employs tempo-invariant encoding of patterns and can thus learn and perform inference on tempo-varying and noisy input data. We demonstrate the model’s capabilities of learning distinctive rhythmic structures of different music genres using unsupervised learning. To test its robustness, we show how the model can efficiently extract rhythmic structures in songs with changing time signatures and live recordings. Additionally, the model’s time-complexity is empirically tested to show its usability for analysis-related applications.
Highlights
Musical rhythm represents one of the basic elements of music
Several different aspects of rhythm and its perception have been explored in the fields of psychology [1], musicology and music theory [2,3], and music information retrieval (MIR) [4]
It is based on relative encoding of time in rhythmic structures, which is commonly used in rhythmic representations and is a necessity, as an individual rhythmic pattern may vary in duration due to tempo changes within a music piece or due to different tempi across pieces in a music corpus
Summary
Musical rhythm represents one of the basic elements of music. Several different aspects of rhythm and its perception have been explored in the fields of psychology [1], musicology and music theory [2,3], and music information retrieval (MIR) [4]. Rhythm is directly related to tempo [5]; rhythm may affect and change the perception of tempo without changing the latter. Rhythmic patterns significantly affect both the melodic and harmonic aspects of a music piece. By changing the underlying rhythmic structure, two versions of a song may be classified into different music genres and imply different dancing styles. Rhythm and its related concepts of tempo and beat have been extensively explored in the music information retrieval field. We present a cross section of related works and provide motivation for developing a new model, presented in this paper
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have