# Masked Contrastive Pre-Training Improves Music Audio Key Detection

arXiv: https://arxiv.org/abs/2604.10021 · HTML: https://mlsanity.com/paper/2604.10021/

## Key facts

- No citing papers in the map yet (data through 7 Oct 2026).
- Builds on 12 of its 21 references that are in the map.
- First posted on arXiv 11 Apr 2026.

## Builds on (most cited first, 12 of 12)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [A Simple Framework for Contrastive Learning of Visual Representations](https://mlsanity.com/paper/2002.05709/) (2020)
- [Scaling Vision Transformers](https://mlsanity.com/paper/2106.04560/) (2021)
- [MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training](https://mlsanity.com/paper/2306.00107/) (2023)
- [Better plain ViT baselines for ImageNet-1k](https://mlsanity.com/paper/2205.01580/) (2022)
- [Contrastive Learning of Musical Representations](https://mlsanity.com/paper/2103.09410/) (2021)
- [Codified audio language modeling learns useful representations for music information retrieval](https://mlsanity.com/paper/2107.05677/) (2021)
- [A Foundation Model for Music Informatics](https://mlsanity.com/paper/2311.03318/) (2023)
- [Supervised and Unsupervised Learning of Audio Representations for Music Understanding](https://mlsanity.com/paper/2210.03799/) (2022)
- [End-to-End Musical Key Estimation Using a Convolutional Neural Network](https://mlsanity.com/paper/1706.02921/) (2017)
- [Myna: Masking-Based Contrastive Learning of Musical Representations](https://mlsanity.com/paper/2502.12511/) (2025)
- [Genre-Agnostic Key Classification With Convolutional Neural Networks](https://mlsanity.com/paper/1808.05340/) (2018)

## Abstract (arXiv)

Self-supervised music foundation models underperform on key detection, which requires pitch-sensitive representations. In this work, we present the first systematic study showing that the design of self-supervised pretraining directly impacts pitch sensitivity, and demonstrate that masked contrastive embeddings uniquely enable state-of-the-art (SOTA) performance in key detection in the supervised setting. First, we discover that linear evaluation after masking-based contrastive pretraining on Mel spectrograms leads to competitive performance on music key detection out of the box. This leads us to train shallow but wide multi-layer perceptrons (MLPs) on features extracted from our base model, leading to SOTA performance without the need for sophisticated data augmentation policies. We further analyze robustness and show empirically that the learned representations naturally encode common augmentations. Our study establishes self-supervised pretraining as an effective approach for pitch-sensitive MIR tasks and provides insights for designing and probing music foundation models.