论文解读
从听见写入选项到用选项做决定：音频大模型中声学证据的两阶段路径
在必须依赖声音的选择题上训练能提升准确率并同步增强对真实音频的行为敏感度，其内部机制是早期到中层先让音频塑造选项表征、再在中层到晚层强化选项对最终答案的贡献，而可学习的 LoRA 更新在模型特定的层带上起决定作用。
注意力机制
论文解读
Word-Level Modeling with Alignment-Aware Acoustic Fusion for Text-Assisted Intelligibility Prediction in Listeners with Hearing Loss
语音质量评估 | 7.7/10
论文解读
Ordering Matters: Rank-Aware Selective Fusion for Blended Emotion Recognition
多模态情感识别 | 5.0/10
论文解读
ChladniSonify: A Visual-Acoustic Mapping Method for Chladni Patterns in New Media Art Creation
音频生成 | 6.0/10
论文解读
Polyphonia: Zero-Shot Timbre Transfer in Polyphonic Music with Acoustic-Informed Attention Calibration
音乐生成 | 7.5/10
论文解读
TARNet: A Temporal-Aware Multi-Scale Architecture for Closed-Set Speaker Identification
TARNet: A Temporal-Aware Multi-Scale Architecture for Closed-Set Speaker Identification
论文解读
When Attention Collapses: Residual Evidence Modeling for Compositional Inference
音频分离 | 7.5/10
论文解读
Pay Attention to CTC: Fast and Robust Pseudo-Labelling for Unified Speech Recognition
语音识别 | 8.0/10
论文解读
A Consistent Learning Depression Detection Framework Integrating Multi-View Attention
语音生物标志物 | 6.5/10
论文解读
A Distribution Matching Approach to Neural Piano Transcription with Optimal Transport
音乐转录 | 7.0/10
论文解读
Adversarial Rivalry Learning for Music Classification
音乐分类 | 6.5/10
论文解读
An Audio-Visual Speech Separation Network with Joint Cross-Attention and Iterative Modeling
语音分离 | 7.5/10
论文解读
Attentive AV-Fusionnet: Audio-Visual Quality Prediction with Hybrid Attention
音视频 | 7.0/10
论文解读
Caption and Audio-Guided Video Representation Learning with Gated Attention for Partially Relevant Video Retrieval
视频检索 | 7.0/10
论文解读
Chunk-Wise Attention Transducers for Fast and Accurate Streaming Speech-to-Text
语音识别 | 7.5/10
论文解读
Combining Multi-Order Attention and Multi-Resolution Discriminator for High-Fidelity Neural Vocoder
语音合成 | 6.5/10
论文解读
DAT-CFTNet: Speech Enhancement for Cochlear Implant Recipients using Attention-based Dual-Path Recurrent Neural Network
语音增强 | 7.0/10
论文解读
Distilling Attention Knowledge for Speaker Verification
说话人验证 | 8.0/10
论文解读
缺失与偏置并存时如何做鲁棒的多模态情感分析：门控序列修复与平衡跨模态注意的协同
📄 缺失与偏置并存时如何做鲁棒的多模态情感分析：门控序列修复与平衡跨模态注意的协同 会议论文 ID：conference:icassp:2026:icassp-arnumber:11460389
论文解读
Expressive Voice Conversion with Controllable Emotional Intensity
语音转换 | 7.5/10