Computational models of hearing and hearing loss

加拿大至少有10%的人口患有听力损失。助听器和人工耳蜗植入提供了一些好处，但言语理解并没有完全恢复，特别是在困难的听力环境中。将开展一些项目，以更好地了解听力损失对语言理解的影响，开发改进的助听器，并了解人工耳蜗的生理功能。第一个目标是进一步发展的计算模型的正常和受损的耳朵。目前的模型能够描述在猫中测量的广泛的生理数据。然而，猫耳和人耳之间存在一些差异，这些差异可能对声音的精确神经表征很重要。因此，我们将创建模型的人类版本，修改模型参数以适应可用的人类数据。耳朵的模型可以使用大脑中的语音处理模型来预测音量和听力损失对人类语音清晰度的影响。然而，该模型没有考虑神经响应的精细定时，这为神经元的平均放电率所传达的信息提供了额外的信息。因此，我们将调查的方法，将尖峰定时信息到我们的语音清晰度预测。该模型也将被应用于开发新的助听器放大方案。我们的调查表明，传统助听器只能平均恢复语音的神经表征，但对于单个语音声音通常无法实现最佳放大。我们将开发语音处理算法，以更优化地恢复所有语音的神经反应。目前的听神经模型无法解释耳蜗植入刺激反应中观察到的许多现象。最近的听神经的生理学研究表明，存在一些在以前的建模研究中没有考虑的离子通道类型。将研究在听觉神经模型中包括这些离子通道类型的影响，将模拟结果与使用人工耳蜗刺激的生理实验数据进行比较。

At least 10% of the population of Canada suffers from hearing loss. Hearing aids and cochlear implants provide some benefit, but speech understanding is not fully restored, particularly in difficult listening environments. A number of projects will be undertaken to better understand the effects of hearing loss on speech understanding, to develop improved hearing aids, and to understand the physiological function of cochlear implants. The first objective is the further development of a computational model of the normal and impaired ear. The current model is able to describe a wide range of physiological data measured in cats. However, there are some differences between the cat and human ear that could be important in the exact neural representation of sounds. Therefore, we will create a human version of the model, with model parameters modified to fit the available human data. The model of the ear can predict the effects of sound volume and hearing loss on human speech intelligibility using a model of speech processing in the brain. However, this model does not take into account the fine timing of neural responses, which provides additional information to that conveyed by the average firing rate of neurons. Consequently, we will investigate methods for incorporating spike-timing information into our speech intelligibility predictor. The model will also be applied to development of novel hearing-aid amplification schemes. Our investigations have shown that conventional hearing aids only restore the neural representation of speech on average, but optimal amplification is often not achieved for individual speech sounds. We will develop speech processing algorithms to more optimally restore neural responses for all speech sounds. Present models of the auditory nerve are unable to explain a number of phenomena observed in responses to cochlear implant stimulation. Recent physiological investigations of the auditory nerve have shown the presence of a number of ionic channel types not considered in previous modeling studies. The effects of including these ionic channel types in a model of the auditory nerve will be investigated, comparing the simulation results to data from physiological experiments using cochlear implant stimulation.