Investigating the Emergent Audio Classification Ability of ASR Foundation Models
This work addresses the problem of enabling ASR models to perform zero-shot audio classification for general and low-resource applications, showing incremental progress by adapting existing models with simple prompts and debiasing.
The study investigated the zero-shot audio classification ability of ASR foundation models like Whisper and MMS, finding that Whisper outperformed existing zero-shot baselines by an average of 9% on 8 datasets without extra training or parameters, with debiasing and model scaling improving performance.
Text and vision foundation models can perform many tasks in a zero-shot setting, a desirable property that enables these systems to be applied in general and low-resource settings. There has been far less work, however, on the zero-shot abilities of ASR foundation models, with these systems typically fine-tuned to specific tasks or constrained to applications that match their training criterion and data annotation. In this work we investigate the ability of Whisper and MMS, ASR foundation models trained primarily for speech recognition, to perform zero-shot audio classification. We use simple template-based text prompts at the decoder and use the resulting decoding probabilities to generate zero-shot predictions. Without training the model on extra data or adding any new parameters, we demonstrate that Whisper shows promising zero-shot classification performance on a range of 8 audio-classification datasets, outperforming the accuracy of existing state-of-the-art zero-shot baselines by an average of 9%. One important step to unlock the emergent ability is debiasing, where a simple unsupervised reweighting method of the class probabilities yields consistent significant performance gains. We further show that performance increases with model size, implying that as ASR foundation models scale up, they may exhibit improved zero-shot performance.