论文解读
End-to-end Listen, Look, Speak and Act
语音对话系统 | 8.5/10
混合专家
论文解读
Adaptive Task-Incremental Learning For Underwater Acoustic Recognition Based on Mixture-of-Experts Adapter
水下声学目标识别 | 7.0/10
论文解读
Dual-Perspective Multimodal Sentiment Analysis with MoE Fusion: Representation Learning via Semantic Resonance and Divergence
多模态情感分析 | 7.0/10
论文解读
Improving Multimodal Brain Encoding Model with Dynamic Subject-Awareness Routing
脑信号编码 | 8.0/10
论文解读
Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence
多模态模型 | 8.5/10
论文解读
Selective Hub Fusion with Modality-Heterogeneous Experts for Multimodal Emotion Recognition
多模态模型 | 6.5/10
论文解读
SURE: Synergistic Uncertainty-Aware Reasoning for Multimodal Emotion Recognition in Conversations
语音情感识别 | 7.5/10
论文解读
SwitchCodec: Adaptive Residual-Expert Sparse Quantization for High-Fidelity Neural Audio Coding
音频生成 | 8.5/10
论文解读
MoVE: Translating Laughter and Tears via Mixture of Vocalization Experts in Speech-to-Speech Translation
这篇论文旨在解决语音到语音翻译（S2ST）系统普遍丢失源语音中非语言声音（如笑声、哭声）和情感信息的问题，这严重影响了跨语言交流的自然度和准确性。为此，作者提出了三项核心贡献：首先，设计了一个可扩展的自动化数据合成管道，用于生成大规模、高质量的英中富有表现力S2ST平行语料，克服了训练数据稀缺的瓶颈