GM-TCNet: Gated Multi-scale Temporal Convolutional Network using Emotion Causality for Speech Emotion Recognition

TL;DR

This paper introduces GM-TCNet, a novel neural network architecture that captures emotional causality in speech using multi-scale gated temporal convolutions, significantly improving speech emotion recognition accuracy.

Contribution

The paper proposes a new gated multi-scale temporal convolutional network with emotional causality learning, enhancing robustness and discriminability in speech emotion recognition.

Findings

Achieves state-of-the-art performance on SER datasets.

Effectively models emotional causality dynamics in speech.

Outperforms existing methods in most evaluation metrics.

Abstract

In human-computer interaction, Speech Emotion Recognition (SER) plays an essential role in understanding the user's intent and improving the interactive experience. While similar sentimental speeches own diverse speaker characteristics but share common antecedents and consequences, an essential challenge for SER is how to produce robust and discriminative representations through causality between speech emotions. In this paper, we propose a Gated Multi-scale Temporal Convolutional Network (GM-TCNet) to construct a novel emotional causality representation learning component with a multi-scale receptive field. GM-TCNet deploys a novel emotional causality representation learning component to capture the dynamics of emotion across the time domain, constructed with dilated causal convolution layer and gating mechanism. Besides, it utilizes skip connection fusing high-level features from different gated convolution blocks to capture abundant and subtle emotion changes in human speech. GM-TCNet first uses a single type of feature, mel-frequency cepstral coefficients, as inputs and then passes them through the gated temporal convolutional module to generate the high-level features. Finally, the features are fed to the emotion classifier to accomplish the SER task. The experimental results show that our model maintains the highest performance in most cases compared to state-of-the-art techniques.