Adapting WavLM for Speech Emotion Recognition

arXiv:2405.04485v1 Announce Type: new
Abstract: Recently, the usage of speech self-supervised models (SSL) for downstream tasks has been drawing a lot of attention. While large pre-trained models commonly outperform smaller models trained from scratch, questions regarding the optimal fine-tuning strategies remain prevalent. In this paper, we explore the fine-tuning strategies of the WavLM Large model for the speech emotion recognition task on the MSP Podcast Corpus. More specifically, we perform a series of experiments focusing on using gender and semantic …

abstract arxiv attention cs.lg cs.sd eess.as emotion explore fine-tuning paper pre-trained models questions recognition scratch speech speech emotion ssl strategies tasks type usage while