用扩散模型填补声学响应空缺，提升麦克风阵列效果
On the Usefulness of Diffusion-Based Room Impulse Response Interpolation to Microphone Array Processing
- 基于扩散模型填补缺失的房间冲激响应
- 实测数据验证方法在真实场景中有效
- 适合做语音增强与空间音频处理的研究者
房间冲激响应（Room Impulse Responses）估计是空间音频处理和语音增强中的基础问题。本文基于此前提出的基于扩散的补全框架，进一步验证其在提升实际多麦克风阵列处理任务性能方面的适用性。同时，在真实世界房间冲激响应数据上测试了该方法的鲁棒性，结果表明其能有效填补缺失信息，显著改善后续音频处理效果。
原文摘要 · Abstract (English)
Room Impulse Responses estimation is a fundamental problem in spatial audio processing and speech enhancement. In this paper, we build upon our previously introduced diffusion-based inpainting framework for Room Impulse Response interpolation and demonstrate its applicability to enhancing the performance of practical multi-microphone array processing tasks. Furthermore, we validate the robustness of this method in interpolating real-world Room Impulse Responses.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。