Translation-Equivariant Self-Supervised Learning for Pitch Estimation with Optimal Transport

2025/08/02 by Torres, Bernardo, Riou, Alain, Richard, Gaël +1
#Artificial Intelligence (cs.AI) #Audio and Speech Processing (eess.AS) #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Sound (cs.SD) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2508.01493

Abstract

In this paper, we propose an Optimal Transport objective for learning one-dimensional translation-equivariant systems and demonstrate its applicability to single pitch estimation. Our method provides a theoretically grounded, more numerically stable, and simpler alternative for training state-of-the-art self-supervised pitch estimators.

Citations

• PESTO: Real-Time Pitch Estimation with Self-supervised Transposition-equivariant Objective
• S-KEY: Self-supervised Learning of Major and Minor Keys from Audio
• STONE: Self-supervised Tonality Estimator
• Toward Fully Self-Supervised Multi-Pitch Estimation
• PESTO: Pitch Estimation With Self-Supervised Transposition-Equivariant Objective
• Cross-domain Neural Pitch and Periodicity Estimation
• Equivariant Self-Supervision for Musical Tempo Estimation
• The Wasserstein-Fourier Distance for Stationary Time Series
• CREPE: A Convolutional Representation for Pitch Estimation
• Optimal spectral transportation with application to music transcription

Related

• PESTO: Pitch Estimation With Self-Supervised Transposition-Equivariant Objective
• PESTO: Real-Time Pitch Estimation with Self-supervised Transposition-equivariant Objective
• Investigating an Overfitting and Degeneration Phenomenon in Self-Supervised Multi-Pitch Estimation
• Toward Fully Self-Supervised Multi-Pitch Estimation
• SLASH: Self-Supervised Speech Pitch Estimation Leveraging DSP-derived Absolute Pitch