Generative Data Augmentation Guided by Triplet Loss for Speech Emotion Recognition

TL;DR

This paper introduces a GAN-based data augmentation method guided by triplet loss to enhance speech emotion recognition, especially under data scarcity and class imbalance conditions, demonstrating significant performance improvements.

Contribution

The paper proposes a novel GAN-guided augmentation approach using triplet loss for SER, effectively addressing data imbalance and cross-lingual challenges.

Findings

Improves recall by 8% on imbalanced datasets

Enhances cross-lingual SER performance with limited target data

Demonstrates effectiveness across multiple languages

Abstract

Speech Emotion Recognition (SER) is crucial for human-computer interaction but still remains a challenging problem because of two major obstacles: data scarcity and imbalance. Many datasets for SER are substantially imbalanced, where data utterances of one class (most often Neutral) are much more frequent than those of other classes. Furthermore, only a few data resources are available for many existing spoken languages. To address these problems, we exploit a GAN-based augmentation model guided by a triplet network, to improve SER performance given imbalanced and insufficient training data. We conduct experiments and demonstrate: 1) With a highly imbalanced dataset, our augmentation strategy significantly improves the SER performance (+8% recall score compared with the baseline). 2) Moreover, in a cross-lingual benchmark, where we train a model with enough source language utterances but very few target language utterances (around 50 in our experiments), our augmentation strategy brings benefits for the SER performance of all three target languages.