Self-Supervised Learning of Context-Aware Pitch Prosody Representations

TL;DR

This paper introduces self-supervised methods to learn context-aware pitch prosody representations from $F_0$ in music and speech, improving downstream classification tasks by up to 15%.

Contribution

It proposes three novel self-supervised paradigms for learning multi-level contextual pitch representations from $F_0$, enhancing MIR tasks.

Findings

Contextual representations improve classification accuracy by up to 15%.

Self-supervised paradigms effectively learn meaningful pitch features.

Representations capture both short-term and long-term contextual information.

Abstract

In music and speech, meaning is derived at multiple levels of context. Affect, for example, can be inferred both by a short sound token and by sonic patterns over a longer temporal window such as an entire recording. In this letter, we focus on inferring meaning from this dichotomy of contexts. We show how contextual representations of short sung vocal lines can be implicitly learned from fundamental frequency ($F_0$) and thus be used as a meaningful feature space for downstream Music Information Retrieval (MIR) tasks. We propose three self-supervised deep learning paradigms which leverage pseudotask learning of these two levels of context to produce latent representation spaces. We evaluate the usefulness of these representations by embedding unseen pitch contours into each space and conducting downstream classification tasks. Our results show that contextual representation can enhance downstream classification by as much as 15\% as compared to using traditional statistical contour features.