ViToSA: Audio-Based Toxic Spans Detection on Vietnamese Speech Utterances

2025/05/31 by H Do, Do, Huy Ba, Vy Le-Phuong Huynh +3
Computer Science · Psychology · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Speech Recognition and Synthesis #Speech and Audio Processing #Stuttering Research and Treatment

paper · pdf · doi:10.48550/arxiv.2506.00636

openalex publication_date 2025/05/31 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Toxic speech on online platforms is a growing concern, impacting user experience and online safety. While text-based toxicity detection is well-studied, audio-based approaches remain underexplored, especially for low-resource languages like Vietnamese. This paper introduces ViToSA (Vietnamese Toxic Spans Audio), the first dataset for toxic spans detection in Vietnamese speech, comprising 11,000 audio samples (25 hours) with accurate human-annotated transcripts. We propose a pipeline that combines ASR and toxic spans detection for fine-grained identification of toxic content. Our experiments show that fine-tuning ASR models on ViToSA significantly reduces WER when transcribing toxic speech, while the text-based toxic spans detection (TSD) models outperform existing baselines. These findings establish a novel benchmark for Vietnamese audio-based toxic spans detection, paving the way for future research in speech content moderation.

Citations

• ViHateT5: Enhancing Hate Speech Detection in Vietnamese With A Unified Text-to-Text Transformer Model
• PhoWhisper: Automatic Speech Recognition for Vietnamese
• VLUE: A New Benchmark and Multi-task Knowledge Transfer Learning for Vietnamese Natural Language Understanding
• MuTox: Universal MUltilingual Audio-based TOXicity Dataset and Zero-shot Detector
• ViSoBERT: A Pre-Trained Language Model for Vietnamese Social Media Text Processing
• Lightweight Toxicity Detection in Spoken Language: A Transformer-based Approach for Edge Devices
• ViHOS: Hate Speech Spans Detection for Vietnamese
• Robust Speech Recognition via Large-Scale Weak Supervision
• DeToxy: A Large-Scale Multimodal Dataset for Toxicity Classification in Spoken Utterances
• Choosing appropriate language to reduce the stigma around mental illness and substance use disorders
• An Exploratory Analysis of the Relation Between Offensive Language and\n Mental Health
• wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
• PhoBERT: Pre-trained language models for Vietnamese
• Unsupervised Cross-lingual Representation Learning at Scale
• DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
• The Sounds of Cyber Threats
• Linguistic Analysis of Toxic Behavior in an Online Video Game
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Related

• MuTox: Universal MUltilingual Audio-based TOXicity Dataset and Zero-shot Detector
• DeToxy: A Large-Scale Multimodal Dataset for Toxicity Classification in Spoken Utterances
• ToxicTone: A Mandarin Audio Dataset Annotated for Toxicity and Toxic Utterance Tonality
• ViHOS: Hate Speech Spans Detection for Vietnamese
• ViHateT5: Enhancing Hate Speech Detection in Vietnamese With A Unified Text-to-Text Transformer Model