Investigating Transformer Encoders and Fusion Strategies for Speech Emotion Recognition in Emergency Call Center Conversations.

Abstract

The emotion detection technology to enhance human decision-making is an important research issue for real-world applications, but real-life emotion datasets are relatively rare and small. The experiments conducted in this paper use the CEMO, which was collected in a French emergency call center. Two pre-trained models based on speech and text were fine-tuned for speech emotion recognition. Using pre-trained Transformer encoders mitigates our data's limited and sparse nature. This paper explores the different fusion strategies of these modality-specific models. In particular, fusions with and without cross-attention mechanisms were tested to gather the most relevant information from both the speech and text encoders. We show that multimodal fusion brings an absolute gain of 4-9% with respect to either single modality and that the Symmetric multi-headed cross-attention mechanism performed better than late classical fusion approaches. Our experiments also suggest that for the real-life CEMO corpus, the audio component encodes more emotive information than the textual one.

The 4th Workshop on Modeling Socio-Emotional and Cognitive Processes from Multimodal Data In-the-Wild (MSECP-Wild)

“…. Deschamps-Berger et al [3] investigate the use of pre-trained and fne-tuned Transformer models for audio and text modalities for emotion recognition. It provides a use-case of how to apply pre-trained machine learning architectures to deal with limited data available in specifc contexts.…”