Quantifying Musical Style: Ranking Symbolic Music based on Similarity to a Style

TL;DR

This paper introduces StyleRank, a computational method for quantifying musical style similarity in MIDI files, which correlates well with human perception and enables scalable, on-demand style comparison.

Contribution

The paper presents a novel feature encoding and Random Forest-based embedding to measure style similarity, improving scalability and interpretability over prior methods.

Findings

StyleRank correlates highly with human perception of musical style.

It can rank generated music samples based on style similarity.

Similarity measurement can be broken down to specific musical features.

Abstract

Modelling human perception of musical similarity is critical for the evaluation of generative music systems, musicological research, and many Music Information Retrieval tasks. Although human similarity judgments are the gold standard, computational analysis is often preferable, since results are often easier to reproduce, and computational methods are much more scalable. Moreover, computation based approaches can be calculated quickly and on demand, which is a prerequisite for use with an online system. We propose StyleRank, a method to measure the similarity between a MIDI file and an arbitrary musical style delineated by a collection of MIDI files. MIDI files are encoded using a novel set of features and an embedding is learned using Random Forests. Experimental evidence demonstrates that StyleRank is highly correlated with human perception of stylistic similarity, and that it is precise enough to rank generated samples based on their similarity to the style of a corpus. In addition, similarity can be measured with respect to a single feature, allowing specific discrepancies between generated samples and a particular musical style to be identified.