面向远场拾音的微分声场测量理论与关键技术研究

Measuring differential sound pressure field (DSPF) aims to make small-aperture microphone arrays have high directivity and frequency-invariant spatial response such that they can capture the signal of desired source from the far-field with high signal-to-noise ratio and high fidelity. Recently, the DSPF measurement has attracted a lot of researchers’ attention due to a wide range of audio applications, such as hands-free speech communication, smart home, and audio recording. Currently, it is the key developmental stage of DSPF measurement. The fundamental theory is clear, but, lack of a systematic analysis; the bottleneck problem is improved, but, not solved. Our research in this project includes three folds. Firstly, we investigate the relationship between the array geometry and the parameters of the DSPF that the array is allowed to measure; This work is aimed to complete the theory of DSPF measurement. Secondly, we systematically study the array self noise, propose an efficient model to characterize it, and propose an approach to reducing its amplification; This work is aimed to solve the bottleneck problem of higher order DSPF measurement. Thirdly, we propose the theory of optimal DSPF based on the noise environment and the array geometry; This work is aimed to help designing the optimal differential microphone arrays. There is no doubt that the results of the project will contribute to theoretical development and engineering implementation of DSPF measurement.

测量微分声场旨在使小孔径麦克风阵列获得高指向和大带宽的空间响应，无需靠近声源，利用空间选择性便能获取高质量的声源信号。这项技术现已成为远场拾音领域的研究热点之一，正在改变未来拾音技术的发展方向，并且拥有广阔的应用前景，诸如通讯设备中实现高质量免提通话，智能家居中实现无感操作体验，以及录音设备中实现分离式录音。微分声场测量技术当前正处在发展的关键时期，测量原理基本清晰，但是缺乏系统的理论；工程瓶颈问题基本定位，并得到改善，但尚未真正解决。针对这一现状，本项目以波域声场理论、噪声精准建模方法和最优化滤波器设计理论为基础，拟开展以下三个方面的研究：①明确可测微分声场与麦克风阵列几何结构之间的约束关系，完善微分声场测量理论；②建立自噪声精准模型，提出抑制方法，更好地解决噪声放大问题；③提出最优微分声场理论及测量方法，为工程应用提供支撑。研究成果必将积极推动微分声场测量技术的发展和工程应用。

测量微分声场旨在使小孔径麦克风阵列获得高指向和大宽带的空间响应，无需靠近声源，利用空间的选择性便能获取高质量的声源信号。这项技术现已成为远场拾音领域的研究热点之一，正在改变未来拾音技术的发展方向，并且拥有广阔的应用前景，诸如通讯设备中实现高质量免提通话，智能家居中实现无感操作体验，以及录音设备中实现分离式录音。针对微分声场测量技术在实际应用中的诸多问题，项目提出了微分声场最优构成方法，提出了最优阵列结构优化方法，提出了基于差分波束的阵列观测模型，提出了非线性、半自适应的空间分布估计方法，提出了利用阵列结构先验的信号参数估算方法。系列成果推动了微分声场测量技术的发展和工程应用。