Pre-Finetuning for Few-Shot Emotional Speech Recognition

TL;DR

This paper introduces a pre-finetuning approach for speech models to improve few-shot emotional speech recognition, enhancing generalization across speakers and domains.

Contribution

It proposes a novel pre-finetuning method for speech models, inspired by NLP transfer learning, to better handle few-shot emotional speech classification tasks.

Findings

Pre-finetuning on diverse corpora improves few-shot recognition accuracy.

The approach reduces speaker overfitting in emotional speech tasks.

Experimental results show significant gains over baseline models.

Abstract

Speech models have long been known to overfit individual speakers for many classification tasks. This leads to poor generalization in settings where the speakers are out-of-domain or out-of-distribution, as is common in production environments. We view speaker adaptation as a few-shot learning problem and propose investigating transfer learning approaches inspired by recent success with pre-trained models in natural language tasks. We propose pre-finetuning speech models on difficult tasks to distill knowledge into few-shot downstream classification objectives. We pre-finetune Wav2Vec2.0 on every permutation of four multiclass emotional speech recognition corpora and evaluate our pre-finetuned models through 33,600 few-shot fine-tuning trials on the Emotional Speech Dataset.