ToxicTone: A Mandarin Audio Dataset Annotated for Toxicity and Toxic Utterance Tonality
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866912385994850304 |
|---|---|
| author | Luo, Yu-Xiang Lin, Yi-Cheng Chuang, Ming-To Chen, Jia-Hung Tsai, I-Ning Kiew, Pei Xing Huang, Yueh-Hsuan Liu, Chien-Feng Chen, Yu-Chen Feng, Bo-Han Ren, Wenze Lee, Hung-yi |
| author_facet | Luo, Yu-Xiang Lin, Yi-Cheng Chuang, Ming-To Chen, Jia-Hung Tsai, I-Ning Kiew, Pei Xing Huang, Yueh-Hsuan Liu, Chien-Feng Chen, Yu-Chen Feng, Bo-Han Ren, Wenze Lee, Hung-yi |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_15773 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ToxicTone: A Mandarin Audio Dataset Annotated for Toxicity and Toxic Utterance Tonality Luo, Yu-Xiang Lin, Yi-Cheng Chuang, Ming-To Chen, Jia-Hung Tsai, I-Ning Kiew, Pei Xing Huang, Yueh-Hsuan Liu, Chien-Feng Chen, Yu-Chen Feng, Bo-Han Ren, Wenze Lee, Hung-yi Audio and Speech Processing Computation and Language Sound 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. |
| title | ToxicTone: A Mandarin Audio Dataset Annotated for Toxicity and Toxic Utterance Tonality |
| topic | Audio and Speech Processing Computation and Language Sound |
| url | https://arxiv.org/abs/2505.15773 |