Semi-Supervised Convolutive NMF for Automatic Piano Transcription

TL;DR

This paper introduces a semi-supervised convolutive NMF method for automatic piano transcription, requiring only one note sample per note, and demonstrates its effectiveness compared to other low-rank techniques.

Contribution

The paper presents a semi-supervised approach using convolutive NMF that reduces the need for extensive labeled data in piano transcription tasks.

Findings

Outperforms other low-rank factorization methods on MAPS dataset

Slightly less accurate than supervised deep learning methods

Faces generalization challenges in practical applications

Abstract

Automatic Music Transcription, which consists in transforming an audio recording of a musical performance into symbolic format, remains a difficult Music Information Retrieval task. In this work, which focuses on piano transcription, we propose a semi-supervised approach using low-rank matrix factorization techniques, in particular Convolutive Nonnegative Matrix Factorization. In the semi-supervised setting, only a single recording of each individual notes is required. We show on the MAPS dataset that the proposed semi-supervised CNMF method performs better than state-of-the-art low-rank factorization techniques and a little worse than supervised deep learning state-of-the-art methods, while however suffering from generalization issues.