PSLA: Improving Audio Tagging with Pretraining, Sampling, Labeling, and Aggregation

TL;DR

This paper introduces PSLA, a set of training techniques that significantly improve audio tagging performance on AudioSet and FSD50K, emphasizing the importance of training strategies alongside model architecture.

Contribution

The paper presents PSLA, a comprehensive collection of training techniques that enhance audio tagging accuracy, demonstrating their effectiveness with EfficientNet models.

Findings

Achieved state-of-the-art mAP of 0.474 on AudioSet with a single model.

Surpassed previous best system with 0.439 mAP using fewer parameters.

Set new record of 0.567 mAP on FSD50K.

Abstract

Audio tagging is an active research area and has a wide range of applications. Since the release of AudioSet, great progress has been made in advancing model performance, which mostly comes from the development of novel model architectures and attention modules. However, we find that appropriate training techniques are equally important for building audio tagging models with AudioSet, but have not received the attention they deserve. To fill the gap, in this work, we present PSLA, a collection of training techniques that can noticeably boost the model accuracy including ImageNet pretraining, balanced sampling, data augmentation, label enhancement, model aggregation and their design choices. By training an EfficientNet with these techniques, we obtain a single model (with 13.6M parameters) and an ensemble model that achieve mean average precision (mAP) scores of 0.444 and 0.474 on AudioSet, respectively, outperforming the previous best system of 0.439 with 81M parameters. In addition, our model also achieves a new state-of-the-art mAP of 0.567 on FSD50K.