Institute for Hearing Technology and Acoustics, RWTH Aachen University, Kopernikusstraße 5, 52074 Aachen, Germany.
Int J Environ Res Public Health. 2021 Dec 29;19(1):324. doi: 10.3390/ijerph19010324.
Head-related transfer functions (HRTFs) play a significant role in modern acoustic experiment designs in the auralization of 3-dimensional virtual acoustic environments. This technique enables us to create close to real-life situations including room-acoustic effects, background noise and multiple sources in a controlled laboratory environment. While adult HRTF databases are widely available to the research community, datasets of children are not. To fill this gap, children aged 5-10 years old were recruited among 1st and 2nd year primary school children in Aachen, Germany. Their HRTFs were measured in the hemi-anechoic chamber with a 5-degree × 5-degree resolution. Special care was taken to reduce artifacts from motion during the measurements by means of fast measurement routines. To complement the HRTF measurements with the anthropometric data needed for individualization methods, a high-resolution 3D-scan of the head and upper torso of each participant was recorded. The HRTF measurement took around 3 min. The children's head movement during that time was larger compared to adult participants in comparable experiments but was generally kept within 5 degrees of rotary and 1 cm of translatory motion. Adult participants only exhibit this range of motion in longer duration measurements. A comparison of the HRTF measurements to the KEMAR artificial head shows that it is not representative of an average child HRTF. Difference can be seen in both the spectrum and in the interaural time delay (ITD) with differences of 70 μs on average and a maximum difference of 138 μs. For both spectrum and ITD, the KEMAR more closely resembles the 95th percentile of range of children's data. This warrants a closer look at using child specific HRTFs in the binaural presentation of virtual acoustic environments in the future.
头相关传递函数（HRTF）在三维虚拟声环境的听觉化中，对现代声学实验设计起着重要作用。该技术使我们能够在受控的实验室环境中创建接近现实生活的情况，包括房间声学效果、背景噪声和多个声源。虽然成人 HRTF 数据库广泛提供给研究界，但儿童数据集却没有。为了填补这一空白，我们在德国亚琛的一年级和二年级小学生中招募了 5-10 岁的儿童。他们的 HRTF 是在半消声室中以 5 度×5 度的分辨率测量的。特别注意通过快速测量程序减少测量过程中运动引起的伪影。为了用个性化方法所需的人体测量数据补充 HRTF 测量，我们对每个参与者的头部和上半身进行了高分辨率的 3D 扫描。HRTF 测量大约需要 3 分钟。与类似实验中的成人参与者相比，儿童在这段时间内的头部运动幅度更大，但通常保持在旋转 5 度和平移 1 厘米的范围内。只有在更长时间的测量中，成年参与者才会表现出这种运动范围。将 HRTF 测量值与 KEMAR 人工头进行比较表明，它不能代表平均儿童 HRTF。在频谱和耳间时间延迟（ITD）方面都存在差异，平均差异为 70 μs，最大差异为 138 μs。对于频谱和 ITD，KEMAR 更接近儿童数据范围的第 95 百分位数。这表明在未来的双耳呈现虚拟声环境中，需要更仔细地考虑使用特定于儿童的 HRTF。
