VoxSafeBench: Not Just What Is Said, but Who, How, and Where
cs.SD, cs.LG, eess.AS
Submitted: 2026-04-16
Updated: 2026-11-16
Code: https://github.com/bitcoin/bips
Project page: https://amphionteam.github.io/VoxSafeBench_demopage
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- MMSU: A Massive Multi-task Spoken Language Understanding and Reasoning Benchmark
- MMAU-Pro: A Challenging and Comprehensive Benchmark for Holistic Evaluation of Audio General Intelligence
- VoiceBench: Benchmarking LLM-Based Voice Assistants
- WavBench: Benchmarking Reasoning, Colloquialism, and Paralinguistics for End-to-End Spoken Dialogue Models
- JALMBench: Benchmarking Jailbreak Vulnerabilities in Audio Language Models
- VoxPrivacy: A Benchmark for Evaluating Interactional Privacy of Speech Language Models
- HearSay Benchmark: Do Audio LLMs Leak What They Hear?
- AudioTrust: Benchmarking the Multifaceted Trustworthiness of Audio Large Language Models
- AHELM: A Holistic Evaluation of Audio-Language Models
- Safety Assessment of Chinese Large Language Models
- SafetyBench: Evaluating the Safety of Large Language Models
- SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal
- TRIDENT: Benchmarking LLM Safety in Finance, Medicine, and Law
- SG-Bench: Evaluating LLM Safety Generalization Across Diverse Tasks and Prompt Types
- A Holistic Approach to Undesired Content Detection in the Real World
- Multi-Turn Jailbreaks Are Simpler Than They Seem
- Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks
- Universal and Transferable Adversarial Attacks on Aligned Language Models
- Now You Hear Me: Audio Narrative Attacks Against Large Audio-Language Models
- Speech-Audio Compositional Attacks on Multimodal LLMs and Their Mitigation with SALMONN-Guard
Related papers
- Speech-DRAME: A Framework for Human-Aligned Benchmarks in Speech Role-Play
- MuLaCover: Flexible Cover Song Generation via Symbolic Tonal Control
- Optimising Neural Speech Codecs for 300bps Communication using Reinforcement Learning
- Few-Shot Open-Set Audio Classification via Transductive Prototype Refinement and Class Logit Enhancement
- Spectral Masking and Interpolation Attack (SMIA): A Black-box Adversarial Attack against Voice Authentication and Anti-Spoofing Systems
- Remix the Timbre: Diffusion-Based Style Transfer Across Polyphonic Stems