Who calls the shots? Rethinking Few-Shot Learning for Audio

TL;DR

This paper investigates the unique challenges of few-shot learning in audio recognition, emphasizing the importance of application-specific approaches due to audio's multi-label nature and properties like polyphony and SNR.

Contribution

It introduces two new datasets for audio few-shot learning and provides audio-specific insights that challenge assumptions from image-based few-shot learning research.

Findings

No single best model or method for all audio scenarios

Support set selection depends on application context

Audio properties significantly influence few-shot learning performance

Abstract

Few-shot learning aims to train models that can recognize novel classes given just a handful of labeled examples, known as the support set. While the field has seen notable advances in recent years, they have often focused on multi-class image classification. Audio, in contrast, is often multi-label due to overlapping sounds, resulting in unique properties such as polyphony and signal-to-noise ratios (SNR). This leads to unanswered questions concerning the impact such audio properties may have on few-shot learning system design, performance, and human-computer interaction, as it is typically up to the user to collect and provide inference-time support set examples. We address these questions through a series of experiments designed to elucidate the answers to these questions. We introduce two novel datasets, FSD-MIX-CLIPS and FSD-MIX-SED, whose programmatic generation allows us to explore these questions systematically. Our experiments lead to audio-specific insights on few-shot learning, some of which are at odds with recent findings in the image domain: there is no best one-size-fits-all model, method, and support set selection criterion. Rather, it depends on the expected application scenario. Our code and data are available at https://github.com/wangyu/rethink-audio-fsl.