ToxicTone: A Mandarin Audio Dataset Annotated for Toxicity and Toxic Utterance Tonality

TL;DR

ToxicTone is a comprehensive Mandarin audio dataset with detailed toxicity annotations, enabling improved detection of toxic speech by leveraging acoustic, linguistic, and emotional cues in spoken language.

Contribution

We created the largest annotated Mandarin toxicity dataset and proposed a multimodal detection framework that outperforms text-only models.

Findings

Our model achieves higher accuracy than baseline models.

Speech cues significantly improve toxicity detection.

The dataset captures diverse real-world communication scenarios.

Abstract

Despite extensive research on toxic speech detection in text, a critical gap remains in handling spoken Mandarin audio. The lack of annotated datasets that capture the unique prosodic cues and culturally specific expressions in Mandarin leaves spoken toxicity underexplored. To address this, we introduce ToxicTone -- the largest public dataset of its kind -- featuring detailed annotations that distinguish both forms of toxicity (e.g., profanity, bullying) and sources of toxicity (e.g., anger, sarcasm, dismissiveness). Our data, sourced from diverse real-world audio and organized into 13 topical categories, mirrors authentic communication scenarios. We also propose a multimodal detection framework that integrates acoustic, linguistic, and emotional features using state-of-the-art speech and emotion encoders. Extensive experiments show our approach outperforms text-only and baseline models, underscoring the essential role of speech-specific cues in revealing hidden toxic expressions.