Complex Pitch Perception in Complex Environments

描述（由申请人提供）：该项目的长期目标是进一步了解听觉系统如何形成音高感知，以及如何使用音高或基频来区分来自不同来源的声音。该项目有三个主要目标，它们都将心理物理实验与计算建模相结合。第一个目的是阐明在正常听力和受损听力中单音复音的音高是如何编码的。我们假设，可分辨谐波的可用性降低，加上单个谐波编码精度的降低，可以预测听力受损听众在更复杂的声学环境中遇到的许多困难。第二个目标是继续寻找我们感知多种音调的能力的机制，并在其他竞争复合体存在的情况下听到一个音调。通过仔细研究受控的情况，刺激要么在频谱上完全重叠，要么根本不重叠，我们能够控制复合体中单个谐波被分解的程度。这些情况为更复杂的情况提供了模型，在这些情况下，多个部分重叠的声音结合在一起，形成了我们日常的声音环境。第三个目标是通过研究音高和基本频率在语音和音乐感知中的作用，采取更实用的方法。通过模拟人工耳蜗的处理过程，我们将阐明在残余听力条件下（声学加电刺激）提供益处的机制，并评估改善音高信息所带来的益处程度。我们还将开始系统地分析在日常声学环境中存在（或缺乏）已解决的谐波，例如在有相互竞争的说话者在场的讲话和小型合奏中的音乐。来自这些分析的数据将为复杂音高感知模型提供强有力的约束，这些模型在不同程度上依赖于已分辨谐波的存在。了解音高在正常和受损听觉系统中是如何编码和利用的，对于听觉假体（如人工耳蜗）的设计和开发以及自动语音识别系统中的源分离算法具有重要意义。

DESCRIPTION (provided by applicant): The long-term goal of this project is to further our understanding of how the auditory system forms pitch percepts and how pitch or fundamental frequency is used to separate sounds coming from different sources. The project has 3 main aims, all of which combine psychophysical experiments with computational modeling. The first aim is to elucidate how the pitch of single harmonic complex sounds is coded in normal and impaired hearing. We hypothesize that the reduced availability of resolved harmonics, in conjunction with a reduction in the coding accuracy of individual harmonics, can predict many difficulties experienced by hearing-impaired listeners in more complex acoustic environments. The second aim is to continue our search for the mechanisms underlying our ability to perceive multiple pitches and to hear out 1 pitch in the presence of other competing complexes. By studying carefully controlled situations where the stimuli either completely overlap in spectrum or do not overlap at all, we are able to control the degree to which individual harmonics within the complexes are resolved. These situations serve as models for the more complex situations, where multiple partially-overlapping sounds combine to form our everyday acoustic environment. The third aim takes a more applied approach by examining the role of pitch and fundamental frequency in speech and music perception. Using simulations of cochlear-implant processing we will elucidate the mechanisms involved in providing a benefit under conditions of residual hearing (acoustic plus electric stimulation) and assess the degree to which the benefit can be ascribed to improved pitch information. We will also begin a systematic analysis of the presence (or lack thereof) of resolved harmonics in everyday acoustic situations, such as speech in the presence of competing talkers and music in small ensembles. The data from these analyses will provide strong constraints for models of complex pitch perception that rely to varying degrees on the presence of resolved harmonics. Understanding how pitch is coded and utilized in the normal and impaired auditory system should have important implications for the design and development of auditory prostheses, such as cochlear implants, and for source segregation algorithms within automatic speech recognition systems.