Audio Playback: Perceptual optimization of sound presentation over spatially distributed loudspeakers

这一项目解决了本研究股的一般挑战，即开发改进和个性化地向听众呈现声音信号的方法。在第二个资助期，我们会处理研究组所关注的三个问题，即“声音呈现”、“干扰噪音”和“个人化”。在这里，我们将考虑调整扬声器输入信号的方法，例如在有混响和干扰噪声源的房间里收听时提高语音清晰度。此外，我们会考虑在量度房间的声学特性的情况下，个别优化声音呈现的方法。在第二个资助期，我们会延长在第一个资助期展开的语音清晰度工作。提出了一种预处理方法，通过最小化房间混响对干语音信号的听觉掩蔽，成功地提高了受房间混响影响的扬声器上再现的语音的可懂度。在下一个资助期，我们会把这项工作的范围扩大，把干扰噪音的听觉掩蔽效应包括在内，以便我们也可以在有干扰噪音的混响房间内提高语音的清晰度。与正在开发语音清晰度预测模型的TP E合作，我们将找到优化频谱内容的算法，并在需要时对听力受损的个人听众应用动态压缩。在第一个资助期，我们还研究了在特定回放房间中记录的声场在感知上的准确再现。正如在听力测试中所展示的，我们能够适应回放房间的各个房间的声学特性。在第二个资助期，我们将从三个方面扩展这项工作。首先，它将扩展到能够通过放置在录音室中听者位置的单个记录单元来记录和再现空间分布的声源。在这里，将寻求与两性生殖系统相结合。为了能够优化混响声场的特性，需要研究混响系统如何激励回放室内的混响声场。第二个努力将与以下事实有关：在当前的优化方法中，同时考虑了录音和回放房间的特性。我们想要调查是否有可能使单一的记录格式可以输入到只知道回放房间的回放系统中。第三个努力将着眼于使用传统的立体声内容(这是当前的记录标准)作为建议的再现方法的输入的方法。

This project addresses the general challenge of this Research Unit to develop methods for an improved and individualized presentation of acoustic signals to listeners. In the second funding period we will address all three problems that the Research Unit focusses on, "sound presentation", "interfering noise", and "individualization".In this project we will focus on the sound presentation over loudspeakers. Here we will consider methods to adapt the input signals for loudspeakers such as to enhance the speech intelligibility when listening in a room with reverberation and interfering noise sources. In addition, we will consider methods to individually optimize the sound presentation taking into account measured acoustical properties of the room.In the second funding period we will extend the work on speech intelligibility that was started in the first funding period. A preprocessing method was developed that successfully enhanced the intelligibility of speech reproduced over loudspeakers affected by room-reverberation by minimizing the auditory masking of the dry speech signal by the room reverberation. In the next funding period, we will extend this work by also including the auditory masking effects of interfering noise such that we can also enhance speech intelligibility in a reverberant room with interfenng noise. In collaboration with TP E which is developing speech intelligibility prediction models we will find algorithms that optimize the spectral content and apply dynamics compression if required for an individual listener with hearing impairment.In the first funding period also the perceptually accurate reproduction of a recorded sound field in a specific playback room was investigated. As demonstrated in listening tests, we were able to adapt to the individual room acoustical properties of the playback room. In the second funding period we will extend this work in three manners. First it will be extended to be able to record and reproduce spatially distributed sound sources with a single recording unit that is placed at the listener's position in the recording room. Here a combination with an ambisonics reproduction system will be pursued. It will require to investigate how the ambisonics system excites the reverberant sound field in the playback room in order to be able to optimize the properties of the reverberant sound field.A second effort will be related to the fact that with the current optimization method both the recording and the playback room properties are taken into account. We want to investigate whether it is possible to make a single recording format that can be input to a playback system that only has knowledge about the playback room.A third effort will look at methods to use conventional stereo content (which is the current recording standard) as an input for the proposed reproduction method.