NOISE REDUCTION FOR COCHLEAR IMPLANTS

描述(申请人提供)：该项目旨在开发降噪算法，以提高人工耳蜗使用者的语音清晰度。第一阶段的工作表明，将定向麦克风阵列与双通道降噪算法结合使用，可以改善人工耳蜗用户在噪声中的语音接收能力。拟议的第二阶段工作将扩大范围，将人工耳蜗植入声学拾取和信号处理链中每一个可能的点上的降噪技术包括在内。这项工作考虑了单麦克风和双麦克风阵列，单通道或双通道预处理算法，以及新颖的通道内压缩机控制算法。这些方法中的每一种都将使用各种方法进行研究，从用于快速系统评估的计算机模拟和自动可理解性预测，到使用便携式实时信号处理进行正式评估的人工耳蜗用户的语音接收测试和听力质量评级。声学测试环境(模拟的或实际的)在混响程度、干扰源的数量和稳定性方面将是真实的。最终的全面研究将结合四种不同的麦克风阵列配置来评估优化的信号处理算法的贡献。这项工作的结果将量化优化的降噪选项对人工耳蜗使用者的好处。最有希望的算法将通过与商业合作伙伴的合同协议在第三阶段工作中开发。人工耳蜗术是恢复重度聋人听力的唯一可行的方法。人工耳蜗术的进展已经到了这样一个程度，许多受试者可以在没有唇读帮助的情况下安静地听懂语言；然而，受试者对背景噪音的敏感性仍然比正常听力收听者差得多。该项目的目标是开发降噪系统，以改善人工耳蜗用户在存在背景噪声的情况下的语音接收。

DESCRIPTION (provided by applicant): This project is aimed at developing noise reduction algorithms for improving the intelligibility of speech for cochlear implant users. Phase 1 work showed that using directional microphone arrays in conjunction with dual-channel noise reduction algorithms improves speech reception in noise for cochlear implant users. The proposed Phase 2 work will expand the scope to include noise reduction techniques at every possible point in the cochlear implant acoustic pick-up and signal processing chain. The work considers single and dual microphone arrays, single- or dual-channel preprocessing algorithms, and novel within-channel compressor- control algorithms. Each of these approaches will be studied using methods that range from computer simulations and automatic intelligibility prediction, for rapid system assessments, to speech-reception testing and listening-quality ratings by cochlear implant users with portable real-time signal processing for formal evaluations. Acoustic test environments (simulated or actual) will be realistic with respect to the degree of reverberation, and the number and stationarity of interference sources. A final comprehensive study will evaluate the contributions of optimized signal-processing algorithms in conjunction with four different microphone-array configurations. The results of this work will quantify the benefits of optimized noise reduction options for cochlear implant users. The most promising algorithms will be developed in Phase 3 work through contractual agreements with commercial partners. Cochlear implants are the only viable procedure for restoring a sense of hearing to profoundly deafened individuals. Progress in cochlear implants has increased to the point where many recipients can understand speech in quiet without the aid of lip-reading; however, susceptibility to background noise continues to be much worse for recipients than for normal hearing listeners. The goal of this project is to develop noise reduction systems to improve speech reception for cochlear implant users in the presence of background noise.