A Direct Brain to Speech Generator for use in Humans

描述（由申请人提供）：长期的竞争力和具体目标：我们的目标是恢复近会话率讲话锁定的个人。在第1阶段研究中，自2004年12月以来植入神经营养电极系统的23岁闭锁受试者的言语运动区的神经记录产生了映射到音素表征以及想象和实际运动的神经数据。在拟议的工作中，我们打算将复杂的语音识别算法，如人工神经网络和隐马尔可夫模型，以实现快速模式识别的实时语音假体的发展。此外，如针对慢性运动研究执行的群体向量分析可以实现将个体神经元放电转换成音素或发音空间以用于驱动语音合成模型的方法。将对另一名患者植入电极系统，以扩展并验证初始受试者完成的工作。由此产生的数据将大大有助于理解语音产生的皮层组织，并加速为闭锁个体开发语音假体。为数据共享目的而开发的网站将得到扩展，并供合作者和其他有关各方使用。 研究与公共卫生的相关性：患有ALS和脑干中风的闭锁患者非常需要创建语音假体设备。在侵入性神经假体研究中进行的大量研究集中在使瘫痪肢体失去的运动功能恢复或通过计算机软件提供间接通信。这项工作是有帮助的锁定患者，但是，这样的患者表示，实时自发语音是一个更理想的最终应用。本研究的目的是开发一种语言修复装置，使用神经营养电极人类皮层记录系统与复杂的模式识别模型和软件。大多数神经假体研究都集中在恢复瘫痪肢体失去的运动功能或通过计算机软件提供间接交流，但我们认为，实时自发语音将是锁定患者更理想的应用。

DESCRIPTION (provided by applicant): LONG TERM OBJECTIVES AND SPECIFIC AIMS: We aim to restore near conversational rate speech in locked-in individuals. In the Phase 1 study, neural recordings from the speech motor area in a 23 year old locked-in subject implanted with the Neurotrophic Electrode System since December 2004 have yielded neural data that have been mapped to phonemic representations and to imagined and actual movements. In the proposed work, we intend to incorporate sophisticated speech recognition algortithms, such as Artificial Neural Networks and Hidden Markov Models, in order to enable rapid pattern recognition for purposes of a real-time Speech Prosthetic development. In addition, Population Vector Analysis as performed for chronic motor studies may realize a method of converting individual neuronal firings into Phonemic or Articulatory Space for driving a Speech Synthesis Model. An additional patient will be implanted with the electrode system to expand and verify the work achieved with the initial subject. The resulting data will add much to understanding the cortical organization of speech production and accelerate the development of a speech prosthetic for locked-in individuals. The website development for data sharing purposes will be expanded and used by the collaborators and other interested parties. RELEVANCE OF RESEARCH TO PUBLIC HEALTH: The creation of a Speech Prosthetic Device is much needed by locked-in patients suffering from ALS and brain stem stroke. The substantial research being performed in invasive neuroprosthetic studies is focused on enabling recovery of lost motor functions in paralyzed limbs or providing indirect communication through computer software. This work is helpful to locked-in patients; however, such patients have indicated that real-time spontaneous speech is a much more desirable final application. The purpose of this research is to develop a speech prosthetic device using the Neurotrophic Electrode Human Cortical Recording system with sophisticated pattern recognition models and software. The majority of neuroprosthetic studies are focused on enabling recovery of lost motor functions in paralyzed limbs or providing indirect communication through computer software, however we believe that real-time spontaneous speech would be much more desirable application to locked-in patients.