Steering Language Model to Stable Speech Emotion Recognition via Contextual Perception and Chain of Thought

TL;DR

This paper introduces C$^2$SER, a novel large-scale audio language model that improves speech emotion recognition stability and accuracy by integrating perceptual modules, chain of thought reasoning, and self-distillation techniques.

Contribution

C$^2$SER is the first to combine contextual perception, chain of thought, and self-distillation for stable and accurate speech emotion recognition in large-scale audio language models.

Findings

C$^2$SER outperforms existing models like Qwen2-Audio and SECap.

It achieves higher stability and precision in emotion recognition.

Extensive experiments validate its effectiveness.

Abstract

Large-scale audio language models (ALMs), such as Qwen2-Audio, are capable of comprehending diverse audio signal, performing audio analysis and generating textual responses. However, in speech emotion recognition (SER), ALMs often suffer from hallucinations, resulting in misclassifications or irrelevant outputs. To address these challenges, we propose C$^2$SER, a novel ALM designed to enhance the stability and accuracy of SER through Contextual perception and Chain of Thought (CoT). C$^2$SER integrates the Whisper encoder for semantic perception and Emotion2Vec-S for acoustic perception, where Emotion2Vec-S extends Emotion2Vec with semi-supervised learning to enhance emotional discrimination. Additionally, C$^2$SER employs a CoT approach, processing SER in a step-by-step manner while leveraging speech content and speaking styles to improve recognition. To further enhance stability, C$^2$SER introduces self-distillation from explicit CoT to implicit CoT, mitigating error accumulation and boosting recognition accuracy. Extensive experiments show that C$^2$SER outperforms existing popular ALMs, such as Qwen2-Audio and SECap, delivering more stable and precise emotion recognition. We release the training code, checkpoints, and test sets to facilitate further research.