Simulation of One ’ s Own Voice in a Two-parameter Model
Simulation of One ’ s Own Voice in a Two-parameter Model
复制标题
在二参数模型中模拟自己的声音
DOI：
10.1121/1.1907278
复制
发表时间：
2014
影响因子：
2.4
通讯作者：
M. Slaney
中科院分区：
文献类型：
--
作者：
Sook Young Won;J. Berger;M. Slaney
众所周知，人们在听到他们录制的唱歌和说话的声音时经常感到不舒服。与正常听力相比，这种对录音的不熟悉是由于不同的传输机制;听一个人的录音只涉及一个单一的空气传导途径，而我们唱歌和说话时听到的声音主要是由于骨传导途径。尽管这是一个众所周知的现象，但研究人员对自己的听力却很少关注，因为它是一个非常复杂的过程，涉及从声带到听觉的多个路径。此外，我们正在研究活人的感知，因此由于伦理原因，增加了进行机械研究的难度。在这项研究中，我们的目标是通过使用图形均衡器的感知实验来测量自己的听力。我们假设，如果受试者通过改变均衡器上的滑块水平来匹配自听和录音的听力，我们可以确定骨传导声音的频谱特性。首先，我们设计了一个均衡器组成的一组峰架滤波器的八个频带。然后，我们进行了两个实验与不同的群体，要求参与者找到最适合自己的歌唱和讲话的声音，通过处理他们的录音的声音均衡器。我们估计传递函数从空气传导到一个人的自己的听力唱歌和说话的声音的基础上选择的均衡器设置。我们观察到受试者内的传递函数相对一致，并且主要具有带通滤波器，广泛放大约300 Hz至1200 Hz。此外，不同性别、不同歌唱经验的被试之间的平均传递函数也呈现出较高的相似性。最后，在进行实验数据简化和验证实验的基础上，我们成功地推导出了自听觉的双参数模型。
It is well known that people often are uncomfortable while hearing their recorded singing and speaking voice. This unfamiliarity with the recorded voice, compared to normal hearing, is due to a different transmission mechanism; listening to one’s recorded voice only involves a single air-conduction pathway, whereas the voice we hear when we sing and speak is largely due to a boneconduction pathway. Despite the well-known phenomenon, one’s own hearing has received less attention among researchers since it is a very complex process involving multiple paths from vocal cords to hearing sensation. Furthermore, we are studying the perception of living humans, thus adding more difficulty to proceed mechanical studies because of an ethical reason. In this study, we aim to measure one’s own hearing through a perceptual experiment using a graphical equalizer. We assume that if a subject matches a self-hearing and a hearing of recorded voice by altering slider levels on the equalizer, we can determine spectral characteristics of bone-conduction sound. First, we design an equalizer consisting of a set of peak and shelf filters for eight frequency bands. Then, we conduct two experiments with different groups as asking participants to find the best fit to their own singing and speech voices by processing their recorded voice on the equalizer. We estimate transfer functions from air conduction to one’s own hearing for both singing and speaking voices based on the chosen equalizer settings. We observe that the transfer functions intra subject are relatively consistent and features mostly band-pass filters, broadly amplifying around 300 Hz to 1200 Hz. Moreover, the averaged transfer functions among subjects also present relatively high degree of similarity regardless of gender and experience level of singing. Finally, we successfully derive a twoparameter model of self-hearing as proceeding experimental data simplification and a validation experiment.