MFHCA: Enhancing Speech Emotion Recognition Via Multi-Spatial Fusion and Hierarchical Cooperative Attention

TL;DR

This paper presents MFHCA, a novel speech emotion recognition method that combines multi-spatial fusion and hierarchical attention to improve accuracy by effectively capturing emotional cues from spectrograms and raw audio.

Contribution

Introduces MFHCA, a new approach integrating multi-spatial fusion and hierarchical cooperative attention for enhanced speech emotion recognition.

Findings

Achieved 2.6% improvement in weighted accuracy on IEMOCAP.

Achieved 1.87% improvement in unweighted accuracy on IEMOCAP.

Demonstrated effectiveness through extensive experiments.

Abstract

Speech emotion recognition is crucial in human-computer interaction, but extracting and using emotional cues from audio poses challenges. This paper introduces MFHCA, a novel method for Speech Emotion Recognition using Multi-Spatial Fusion and Hierarchical Cooperative Attention on spectrograms and raw audio. We employ the Multi-Spatial Fusion module (MF) to efficiently identify emotion-related spectrogram regions and integrate Hubert features for higher-level acoustic information. Our approach also includes a Hierarchical Cooperative Attention module (HCA) to merge features from various auditory levels. We evaluate our method on the IEMOCAP dataset and achieve 2.6\% and 1.87\% improvements on the weighted accuracy and unweighted accuracy, respectively. Extensive experiments demonstrate the effectiveness of the proposed method.