人工耳蜗言语处理功能的移动端模块化实现

发布时间：2018-05-18 20:12

本文选题：人工耳蜗 + 言语处理策略　；参考：《深圳大学》2017年硕士论文

【摘要】：人工耳蜗(cochlear implant,CI),又称电子耳蜗,是目前世界上最成功的用于人体的神经假体之一。它通过电刺激来代替声刺激,帮助重度/极重度耳聋患者恢复听觉,是目前治疗全聋患者唯一有效的方法。CI产品包括体外机和(体内)植入体两部分。体外机起到信号采集、处理与传输的作用,特别是其中的言语处理模块是影响CI植入者言语辨识能力的关键因素。现有的体外机受其计算性能、功耗、体积等因素的限制,一方面无法提供高效但复杂的言语处理算法所需的计算资源,从而无法提升植入者在复杂噪声环境下的言语辨识能力;另一方面需频繁更换电池和较大的体积也给植入者带来很大的使用不便,甚至心理压力。这限制了CI产品的更大规模应用。本论文针对CI体外机的上述缺点提出一种改进方案:取消体外机的信号采集和言语处理模块,将这部分功能移植到移动终端上实现;体外机中只包含信号传输模块,包括新增的蓝牙模块和原射频模块,分别用于体外机与移动终端和植入体间的信号传输。论文在Android手机系统上实现了该方案:通过手机自带麦克风采集语音信号,利用手机的计算资源进行信号处理(本文采用目前应用面最广的CIS言语编码策略),通过手机自带蓝牙模块将处理后的电信号发送到体外机的蓝牙接收模块,再经过射频传输发送到植入体。利用现有移动终端的高计算性能,不仅可以显著提升已有算法的运算速度,而且为将来运算量更大,处理效果更好的复杂算法提供了实施的空间。另外,本论文针对CI植入者术后康复训练的需要设计了一个辅助训练模块,该模块具有基于语音识别的语音-文字转换功能和语音重复播放功能。耳聋患者在接受CI植入后需要长期的专业言语康复训练才能达到较好的听觉恢复效果,耗时耗资都很昂贵,这也是很多耳聋患者放弃CI植入的因素之一。本文所设计的辅助训练模块有助于植入者在专业康复训练之余通过文字提示和重复听辩进行辅助训练,更快更好地恢复/重建言语听觉能力。最后,本论文设计了一系列的分析实验和听觉仿真实验,验证了所设计系统的可靠性。
[Abstract]:Cochlear implant (CI), also known as the electronic cochlea, is one of the most successful artificial neural prostheses in the world. It uses electrical stimulation instead of acoustic stimulation to help the severe / severely deafened patients recover their hearing. It is the only effective way to treat all deafness patients..CI products include two parts of the body and the implant. In vitro, it plays the role of signal acquisition, processing and transmission, especially the speech processing module is the key factor affecting the speech recognition ability of CI implants. The existing external machine is limited by its computing performance, power consumption, volume and other factors. On the one hand, it can not provide the computing resources needed for efficient but complex speech processing algorithms. It can not improve the speech recognition ability of the implants in the complex noise environment; on the other hand, the frequent replacement of the battery and the larger volume also bring great inconvenience to the implants and even the psychological pressure. This restricts the more large-scale application of the CI products. This paper proposes an improved scheme for the shortcomings of the CI extracorporeal machine: cancellation The signal acquisition and speech processing module of the extracorporeal machine transplants the function to the mobile terminal. In vitro, only the signal transmission module is included, including the new Bluetooth module and the original RF module, which are used for the signal transmission between the external machine and the mobile terminal and the implants respectively. The paper implements this scheme on the Android mobile phone system. Using the mobile phone to collect the voice signal with the microphone and use the computing resources of the mobile phone to carry on the signal processing (this paper uses the most widely used CIS speech coding strategy at present), sends the processed electrical signal to the Bluetooth receiving module of the external machine through the Bluetooth module with the mobile phone, and then sends it to the implant through the radio frequency transmission. The high computing performance of the mobile terminal can not only significantly improve the computing speed of the existing algorithms, but also provide the implementation space for the complex algorithms with better computational complexity and better processing effect. In addition, a supplementary training module is designed for the needs of postoperative rehabilitation training for CI implants. This module has a speech recognition based on speech recognition. The deaf patients need long-term professional speech rehabilitation training to achieve better hearing recovery after CI implantation. It is also one of the factors that a lot of deafness patients give up CI implantation. The auxiliary training module designed in this paper helps to cultivate the hearing loss. At the end of this paper, a series of analytical and auditory simulation experiments have been designed to verify the reliability of the designed system.
【学位授予单位】：深圳大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN912.3

【相似文献】