论文解读
Detecting and Attributing Synthetic Spanish Speech: The HISPASpoof Dataset
语音伪造检测 | 7.5/10
语音伪造检测
论文解读
Disentangled Authenticity Representation for Partially Deepfake Audio Localization
音频深度伪造检测 | 6.5/10
论文解读
EchoFake: A Replay-Aware Dataset For Practical Speech Deepfake Detection
音频深度伪造检测 | 8.5/10
论文解读
Fake Speech Wild: Detecting Deepfake Speech on Social Media Platform
语音伪造检测 | 7.0/10
论文解读
Fine-Grained Frame Modeling in Multi-Head Self-Attention for Speech Deepfake Detection
语音伪造检测 | 8.0/10
论文解读
Hybrid Pruning: In-Situ Compression of Self-Supervised Speech Models for Speaker Verification and Anti-Spoofing
说话人验证 | 8.0/10
会议任务专题
ICASSP 2026 - 语音伪造检测
共 8 篇 ICASSP 2026 语音伪造检测 方向论文
论文解读
Mind Your [m]S, Cross Your [t]S: a Large-Scale Phonetic Analysis of Speech Reproduction in Modern Speech Generators
语音伪造检测 | 7.0/10
论文解读
Multi-Task Transformer for Explainable Speech Deepfake Detection via Formant Modeling
语音伪造检测 | 7.5/10
论文解读
Speech Quality-Based Localization of Low-Quality Speech and Text-to-Speech Synthesis Artefacts
语音质量评估 | 7.0/10
论文解读
Tri-Attention Fusion: Joint Temporal-Spectral and Bidirectional Modeling for Speech Spoofing Detection
语音伪造检测 | 7.0/10
论文解读
WaveSP-Net: Learnable Wavelet-Domain Sparse Prompt Tuning for Speech Deepfake Detection
语音伪造检测 | 8.0/10
论文解读
RTCFake: Speech Deepfake Detection in Real-Time Communication
语音伪造检测 | 7.0/10
论文解读
Spectro-Temporal Modulation Representation Framework for Human-Imitated Speech Detection
语音伪造检测 | 6.5/10
论文解读
Neural Encoding Detection is Not All You Need for Synthetic Speech Detection
这篇综述论文的核心贡献在于**揭示并论证了当前合成语音检测领域的一个关键误区：过度依赖“神经编码检测”**。论文首先系统回顾了基于SincNet、自监督学习（SSL）和神经编码检测的三类数据驱动方法，指出当前性能最佳的SSL模型实际上主要捕捉的是声码器（vocoder）在波形生成阶段引入的痕迹，而非
论文解读
Listening Deepfake Detection: A New Perspective Beyond Speaking-Centric Forgery Analysis
本文首次提出了“聆听深度伪造检测”这一新任务，旨在识别视频中人物在倾听状态下（非说话时）的伪造反应，弥补了现有研究主要集中于“说话”场景的不足。为解决此任务数据稀缺的问题，作者构建了首个专门数据集Li
论文解读
ProSDD: Learning Prosodic Representations for Speech Deepfake Detection against Expressive and Emotional Attacks
这篇论文旨在解决当前语音深度伪造检测（SDD）系统在面对富有表现力和情感的合成语音攻击时泛化能力不足的核心问题。现有方法过度依赖伪造数据，容易学习数据集特定的伪影，而非自然语音的可迁移特征。为此，作者