论文解读
有害不在一处：当声音与画面由多模态共同拼出
音视频生成 | 7.3/10
基准测试
论文解读
不只用文字指挥检索：当嵌入器学会看区域、听时段
音频检索 | 7.5/10
论文解读
说出来就变了味：当语音不再是文本的透明通道
音视频问答 | 8.1/10
论文解读
听错一个字就全盘皆输：SpeechGym 把语音智能体的失败变成可训练的梯度
语音交互 | 7.6/10
论文解读
边播边改：当音视频生成从片段走向持续世界
音频生成 | 8.2/10
论文解读
当文字说得太漂亮，声音却在摇头：如何让模型听出言外之意
语音交互 | 6.6/10
每日研究速递
语音/音乐/音频论文速递 2026-08-29
共分析 29 篇语音/AI 论文
论文解读
ASR 评测别急着换大模型：先分清向量度量、候选裁判与错误解释
语音质量评估 | 6.1/10
论文解读
把 100 小时 MEG 拆成两种稀缺资源，才看得见神经语音解码该怎样扩展
基准测试 | 8.9/10
论文解读
把“该不该开口”拆成可追责时间表：TurnBench 逼出的轮次交接难题
语音交互 | 8.5/10
论文解读
别再把同步分数当裁判：先问它量的是偏移、内容，还是感知代理
音视频理解 | 8.8/10
论文解读
Benchmarking LLM Judges for Voice-Agent Evaluation: Reliability, Calibration, and Human Oversight
语音交互 | 6.6/10
论文解读
EXAM²: Extending Audio Understanding in Multilingual and Multimodal Analysis
音频理解 | 8.7/10
论文解读
Lost in Speech: Trilingual Spoken Hallucination Detection Across Audio and Transcripts
音频理解 | 8.2/10
论文解读
OmniJudge or OmniBias? Diagnosing Multimodal Judges through Balanced, Decoupled Lenses
音频质量评估 | 7.4/10
论文解读
The ISCSLP 2026 Real-World Audio-Visual Speech Enhancement Challenge
音视频语音分离 | 8.8/10
论文解读
AT-ADD: A Benchmark and Challenge for Robust and All-Type Audio Deepfake Detection
音频伪造检测 | 9.6/10
论文解读
MRMAD: A Multi-Round Multi-Audio Benchmark for Evaluating Acoustic Degradation Perception in Large Audio-Language Models
音频质量评估 | 8.9/10
论文解读
Hear2Act: Benchmarking When Prosody Should Change What an Assistant Does
语音交互 | 6.4/10
论文解读
Towards Quantifying Benchmark Optimization in ASR Models
语音识别 | 8.5/10