Models for auditory perception, audio quality and speech intelligibility

听觉感知、音频质量和语音清晰度模型

• 批准号: 203313112
• 负责人: Dr. Stephan Ewert, since 7/2016
• 金额: --
• 依托单位: Abteilung Medizinische Physik
• 依托单位国家: 德国
• 项目类别: Research Units
• 财政年份: 2012
• 资助国家: 德国
• 起止时间: 2011-12-31 至 2019-12-31
• 项目状态: 已结题
• 来源: https://gepris.dfg.de/gepris/projekt/203313112?language=en
• 关键词: Models; auditory; perception; audio; quality

Although the "effective" signal processing for monaural and binaural hearing is relatively well described by numerical models for normal hearing (and to a lesser extent for hearing impaired), there are only few approaches for the practically relevant knowledge transfer to prediction of perceived audio quality and speech intelligibility. This is particularly the case for binaural hearing, which is of importance as many potential applications in hearing acoustics are binaural (e.g. listening quality in rooms, for hearing supportive devices, and for noise exposure at the work place). The goal of the current project is the development and evaluation of audio quality and speech intelligibility models which are based on monaural and binaural perception models. In the context of the Research Unit it is of particular interest that models are suited to account for arbitrary listening (presentation) scenarios for normal heanng as well as for miid-to-moderately hearing impaired (individualization), tackling the presentation problem and the individualization problem. Speech and audio quality for a wide variety of signal distortions are of particular interest as, on the one hand, devices with very high and even transparent characteristics become available for home and mobile applications, and on the other hand, next to (limited) improvement of speech intelligibility (e,g. for single channel noise reduction) listening effort and perceived quality are candidates for algorithm improvement. The models developed in this project will be used by the other projects of the Research Unit for optimization of the respective signal processing algorithms. Moreover, the models will be applied in addition to subjective listening tests for the evaluation of the signal processing systems in the other projects, while subjective data provided by the other projects will be used here to further develop the models.

虽然单耳和双耳听力的“有效”信号处理相对较好地描述了正常听力的数值模型（听力受损程度较低），但只有很少的方法用于实际相关的知识转移到感知音频质量和语音清晰度的预测。这对于双耳听觉尤其如此，这是重要的，因为听觉声学中的许多潜在应用都是双耳的（例如，房间中的收听质量，用于听觉支持设备，以及用于工作场所的噪声暴露）。目前的项目的目标是开发和评估的音频质量和语音清晰度模型的基础上单耳和双耳感知模型。在研究单位的背景下，特别感兴趣的是，模型适合于考虑正常听力的任意听力（呈现）场景以及中度至中度听力受损（个性化），解决呈现问题和个性化问题。对于各种各样的信号失真的语音和音频质量是特别感兴趣的，因为一方面，具有非常高的甚至透明的特性的设备变得可用于家庭和移动的应用，并且另一方面，仅次于（有限的）语音可懂度（例如，对于单信道降噪）收听努力和感知质量是算法改进的候选。在这个项目中开发的模型将用于研究股的其他项目，以优化各自的信号处理算法。此外，除了主观听力测试外，这些模型还将用于评估其他项目中的信号处理系统，而其他项目提供的主观数据将用于进一步开发模型。