SBIR Phase II: (IT-B5) Feasibility to run novel voice interface on a low-power microcontroller

这项小型企业创新研究(SBIR)第二阶段研究项目将实施一个适合视障用户访问要求的微型信息管理系统。目前为视障人士提供的信息技术速度慢，难以在手持白色拐杖或导盲犬的情况下操作。视障人士将从操作更快、更容易访问的免提/免眼信息系统中受益匪浅。该项目将开发一种语音操作的个人数字助理(PDA)，名为Viyan，其执行速度是盲文PDA的10倍。第一阶段研究的结果证明了在没有硬件浮点运算的集成微控制器上实现实时语音处理算法的可行性。这个第二阶段项目的结果预计将导致一种类似于最先进的媒体播放器的可穿戴设备，处理速度快10倍，尺寸小10倍。全球有超过1.6亿视障人士，仅美国就有1000万人，这项拟议的研究是朝着一种设备迈出的关键一步，这种设备将比目前的替代设备更好地满足他们的移动信息管理需求。此外，这款设备应该会对有视力的人的移动信息管理产生影响。在第一阶段项目期间对讲多种语言的视力用户进行的可用性试验结果表明，为视力和视障用户提供强大和快速的图形用户界面替代人机界面的结果。此外，这种语音技术适合集成到移动设备中，例如2007年售出超过3亿台的移动电话。

This Small Business Innovation Research (SBIR) Phase II research project will implement a miniature information management system that is suitable to the access requirements for visually impaired users. Current information technologies for the visually impaired are slow and difficult to operate while holding a white cane or guide dog. Visually impaired people will benefit greatly from a hands-free/eyes-free information system that is much faster to operate and easier to access. The project will develop a voice-operated personal digital assistant (PDA), called Vivian, which performs 10 times faster than Braille PDAs. The outcome of the Phase I study demonstrated the feasibility of real-time speech processing algorithms on integrated microcontrollers without hardware floating-point arithmetic. The outcome of this Phase II project is anticipated to result in a wearable device similar to a state of the art media player with 10X faster processing and 10X smaller in size.With more than 160 million visually impaired people worldwide, 10 million in the US alone, the proposed research is a critical step towards a device that will address their mobile information management needs significantly better than current alternatives. Moreover, this device should impact mobile information management for sighted people. The results of usability trials with sighted users speaking multiple languages conducted during the Phase I project, indicate that the outcomes of a powerful and fast alternative human computer interface to graphical user interfaces for sighted and visually impaired users. Additionally, this voice technology is suitable for integration into mobile appliances such as mobile phones for which over 300 million were sold in 2007.