Associated dataset for "Effects of Additive Noise in Binaural Rendering of Spherical Microphone Array Signals"

The instrumental evaluation utilized the Real-Time Spherical Microphone Renderer (ReTiSAR) for binaural reproduction in Python. The employed code state at that time should be used to reproduce the rendering results in this data set exactly. The frozen code state for this data set is available at: https://github.com/AppliedAcousticsChalmers/ReTiSAR/releases/tag/v2021.TASLP Download the rendering pipeline and follow the setup instructions! Use the Conda environment file included here when setting up the Python environment. In this way, you will obtain the exact Python setup as utilized in the instrumental evaluation in the publication: conda env create --file ReTiSAR_environment_freeze.yml source activate ReTiSAR_TASLP_freeze Matlab script "generate_norm_levels.m": The level contributions used in the publication are contained in "record_CLL_levels.sh", therefore this needs to be executed only in case other level distributions should be generated Generate the string for a ReTiSAR configuration (as used in "record_CLL_levels.sh") to emulate normally contributing EM32 self-noise based on Forum Acusticum publication data Generate the string for aReTiSAR configuration (as used in "record_CLL_levels.sh") to emulate normally contributing GL162 self-noise based on Gaussian normal distribution Matlab script "prepare_MagLS_HRIRs.m": Apply Magnitude Least Squares pre-processing to HRIRs (as used in "record_CLL_levels.sh") Shell script "record_CLL_levels.sh": Record the output ear signals of the rendering pipeline at multiple head orientations for all investigated configurations (according to Table 1) All captured signals are contained in the respective configuration directory, e.g. "rec_25ch_Fliege_sh4" to "rec_338ch_Gauss_sh12" Matlab script "calculate_CLL_levels.m": Read the individual rendering pipeline output recordings for arbitrary configurations Visualize the raw captured signals per configuration Visualize the RMS signal level variations over all head orientations per configuration Visualize the Interaural Level Difference variations over all head orientations per configuration Visualize the Composite Loudness Level variations over all head orientations per configuration (like Figure 2) Gather the above determined RMS, ILD, CLL, etc. metrics in a Matlab dataset per configuration (will be utilized in "plot_gathered_CLL_levels.m") All generated plots and Matlab datasets are contained in the respective configuration directory, e.g. "rec_25ch_Fliege_sh4" to "rec_338ch_Gauss_sh12" Matlab script "plot_gathered_CLL_levels.m": Visualize the resulting Composite Loudness Level gradient detections over all head orientations for arbitrary combinations of configurations (like Figure 3 to Figure 14) All generated plots are contained in the "CLL_results" directory本实验采用Python环境下的实时球面麦克风渲染器（Real-Time Spherical Microphone Renderer，ReTiSAR）进行双耳声重放。需使用本数据集对应的代码版本，方可精准复现本次渲染结果。本数据集对应的固定代码版本已上传至：https://github.com/AppliedAcousticsChalmers/ReTiSAR/releases/tag/v2021.TASLP，请下载该渲染流水线并按照设置说明完成配置。配置Python环境时，请使用本文附带的Conda环境配置文件，具体命令如下： conda env create --file ReTiSAR_environment_freeze.yml source activate ReTiSAR_TASLP_freeze 如此即可得到与本论文实验中完全一致的Python运行环境。 MATLAB脚本"generate_norm_levels.m"： 本论文中使用的声级贡献参数存储于"record_CLL_levels.sh"中，仅当需要生成其他声级分布时，才需执行该脚本。 基于Forum Acusticum会议论文的公开数据，生成用于模拟EM32本底噪声真实贡献特性的ReTiSAR配置字符串（使用方式详见"record_CLL_levels.sh"）。 基于高斯正态分布，生成用于模拟GL162本底噪声的ReTiSAR配置字符串（使用方式详见"record_CLL_levels.sh"）。 MATLAB脚本"prepare_MagLS_HRIRs.m"： 对头部相关冲激响应（Head-Related Impulse Responses，HRIRs）应用幅度最小二乘（Magnitude Least Squares，MagLS）预处理，处理逻辑与"record_CLL_levels.sh"中保持一致。 Shell脚本"record_CLL_levels.sh"： 针对所有待研究的配置（详见表1），在多个头部方位角下录制渲染流水线的输出耳信号。所有录制得到的信号均存储于对应配置目录下，例如"rec_25ch_Fliege_sh4"至"rec_338ch_Gauss_sh12"。 MATLAB脚本"calculate_CLL_levels.m"： 读取任意配置对应的渲染流水线输出录制文件； 可视化各配置下录制的原始耳信号； 可视化各配置下所有头部方位角对应的RMS信号声级变化； 可视化各配置下所有头部方位角对应的耳间声级差（Interaural Level Difference，ILD）变化； 可视化各配置下所有头部方位角对应的复合响度级（Composite Loudness Level，CLL）变化（对应论文中图2）； 将上述得到的RMS、ILD、CLL等指标汇总至各配置对应的MATLAB数据集（将被"plot_gathered_CLL_levels.m"调用）。 所有生成的图表与MATLAB数据集均存储于对应配置目录下，例如"rec_25ch_Fliege_sh4"至"rec_338ch_Gauss_sh12"。 MATLAB脚本"plot_gathered_CLL_levels.m"： 可视化任意配置组合下所有头部方位角对应的复合响度级梯度检测结果（对应论文中图3至图14）。所有生成的图表均存储于"CLL_results"目录下。

搜集汇总

背景与挑战

背景概述

该数据集是研究论文'Effects of Additive Noise in Binaural Rendering of Spherical Microphone Array Signals'的关联数据集，主要用于分析球形麦克风阵列信号在双耳渲染中加性噪声的影响。它包含使用ReTiSAR渲染工具生成的配置文件和脚本，支持声级、双耳声级差和复合响度级等指标的评估，并提供了可复现实验结果的冻结代码状态。