| 题名 | 面向人工耳蜗语音信号处理的ASIC设计研究 |
| 作者 | 何星 |
| 学位类别 | 博士 |
| 答辩日期 | 2008-05-30 |
| 授予单位 | 中国科学院声学研究所 |
| 授予地点 | 声学研究所 |
| 关键词 | 人工耳蜗 语音信号处理 连续间隔采样 低功耗 |
| 其他题名 | ASIC Design and Research of CI Speech Processing |
| 学位专业 | 信号与信息处理 |
| 中文摘要 | 人工耳蜗是目前唯一能使全聋患者恢复听觉的装置，它是向功能尚完好的听神经施加脉冲电刺激来传递声音的。它可以分为体外部分和体内部分，而体外部分的语音信号处理器是人工耳蜗的核心部分。本论文通过分析多种人工耳蜗语音信号处理方案，选取CIS方案进行优化和ASIC设计，以满足人工耳蜗产品低功耗、高性能的需求。 为了降低人工耳蜗系统的功耗，采用ASIC方法来设计其语音信号处理部分，来代替传统的DSP实现形式。算法中的多项参数均可调节，具有较大的灵活性，并且所设计的结构可以适用多种n-of-m类方案。由于提供了较高的刺激速率，并且通过提取基音频率信息来增强汉语的声调信息，从而提高了患者的语音识别率。考虑到仿真过程对于评价人工耳蜗性能的重要性，本文提出了一种更加接近人工耳蜗使用者试听效果的仿真模型。 采用牛顿迭代算法进行CIS方案中对数/指数压缩函数的设计，可以有效地控制查找表的大小，并且具有较快的收敛速度，计算精度和迭代次数均可调。由于在许多场合需要在频域进行语音信号处理，所以本文设计了一个流水线结构的FFT处理单元，具有处理速度快、资源利用高效的特点。 针对体内接受刺激器设计的具体要求，进行了完整的模块划分。为了降低通道间的相互干扰，设计了双相脉冲和三相脉冲生成电路。输出电极开关网络可以提供多种刺激模式。并且可以根据虚拟通道理论，提供给患者更多的频域信息。 在本文的最后，为了进一步降低人工耳蜗系统的功耗，首先分析了系统功耗的构成，并对多种功耗优化方法进行分析。在此基础上，在系统级和RTL级采用多种优化方法降低系统的功耗，为全植入式人工耳蜗的设计提供了依据。 |
| 英文摘要 | Cochlear implant (CI) is currently the only available medical device to restore hearing ability to totally deaf patients. It delivers auditory signals through electrically stimulating the auditory nerve that is functionally well. The cochlear implant consists of external and internal parts. Speech processing of the external part is the core of CI. This dissertation analyzed a variety of speech processing strategies of CI, chose CIS strategy and optimized and designed it to satisfy the requirements of low power consumption and high performance. In order to reduce the power consumption of CI system, this dissertation adopts ASIC method to design speech processor instead of traditional DSP form. Because many parameters of the algorithm can be adjusted, the algorithm is very flexible and can be applied to various n-of-m strategies. By providing a high rate of stimulation and picking up fundamental frequency to enhance the tonal information of Chinese language, it improves the speech recognition rate of patients. Considering the importance of simulation process in evaluating the performance of CI, this dissertation put forward a simulation model that is far closer to the listening effect of CI users. It adopts Newton-Raphson iteration to design two compression functions of logarithm / exponential functions so as to effectively control the size of lookup table, beside, the convergence speed is fast. As speech processing has to be conducted in the frequency domain on many occasions, this dissertation designed a pipelined FFT/IFFT processing unit which has a fast processing speed and high resource utilization efficiency. Aimed at the design requirements of receiver-stimulator, it divides the modules completely. In order to reduce the mutual interference of channels, it designs biphasic-pulse generation circuit and triphasic-pulse generation circuit. Output electrode switching networks can provide a wide range of stimulating modes. And under virtual channel theory, it can provide patients with more frequency-domain information. Finally, in order to further decrease the power consumption of CI system, the dissertation analyzes the source of power consumption; based on this, it then analyzes many power consumption optimization methods at different levels. On this basis, it adopts several specific ways at the system level and the RTL to further reduce the power consumption of the system, which offers a basis for the design of full implants. |
| 语种 | 中文 |
| 公开日期 | 2011-05-07 |
| 页码 | 127 |
| 内容类型 | 学位论文 |
| 源URL | [http://159.226.59.140/handle/311008/306]  |
| 专题 | 声学研究所_声学所博硕士学位论文_1981-2009博硕士学位论文 |
| 推荐引用方式 GB/T 7714 | 何星. 面向人工耳蜗语音信号处理的ASIC设计研究[D]. 声学研究所. 中国科学院声学研究所. 2008. |
| 个性服务 |
| 查看访问统计 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论