An Investigation Into Explainable Audio Hate Speech Detection

TL;DR

This paper introduces the task of explainable audio hate speech detection, proposing two methods, creating a synthetic dataset, and demonstrating that end-to-end models with rationales improve detection accuracy.

Contribution

It presents the first approaches for explainable audio hate speech detection, including a synthetic dataset and two model architectures, with the E2E method showing superior performance.

Findings

E2E approach outperforms cascading in IoU metric

Including frame-level rationales improves detection accuracy

Synthetic dataset enables training of explainable models

Abstract

Research on hate speech has predominantly revolved around detection and interpretation from textual inputs, leaving verbal content largely unexplored. While there has been limited exploration into hate speech detection within verbal acoustic speech inputs, the aspect of interpretability has been overlooked. Therefore, we introduce a new task of explainable audio hate speech detection. Specifically, we aim to identify the precise time intervals, referred to as audio frame-level rationales, which serve as evidence for hate speech classification. Towards this end, we propose two different approaches: cascading and End-to-End (E2E). The cascading approach initially converts audio to transcripts, identifies hate speech within these transcripts, and subsequently locates the corresponding audio time frames. Conversely, the E2E approach processes audio utterances directly, which allows it to pinpoint hate speech within specific time frames. Additionally, due to the lack of explainable audio hate speech datasets that include audio frame-level rationales, we curated a synthetic audio dataset to train our models. We further validated these models on actual human speech utterances and found that the E2E approach outperforms the cascading method in terms of the audio frame Intersection over Union (IoU) metric. Furthermore, we observed that including frame-level rationales significantly enhances hate speech detection accuracy for the E2E approach. \textbf{Disclaimer} The reader may encounter content of an offensive or hateful nature. However, given the nature of the work, this cannot be avoided.