Generic community-powered meta speech recognition platform

• Main Author: M. Mugunth Kumar
• Other Authors: Theng Yin Leng
• Format: Theses and Dissertations
• Language: English
• Published: 2010
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: http://hdl.handle.net/10356/41796
• Institution: Nanyang Technological University

In today's world, for speech enabling applications, a main problem faced by researchers is having to rework from the start or "re-invent the wheel" owing to various reasons like existing speech recognition engines being too inefficient or inaccurate, or that existing speech-enabled applications are simply too specific for consideration. To alleviate this problem, one approach that can be used is a platform-based approach, engaging community of programmers to write applications for this platform. This dissertation, hence, aims at developing a community-supported, open and generic platform allowing other speech-enabled windows based applications to be developed using this platform. The dissertation aims at bridging two disparate areas of computer science, namely speech recognition and automation. Speech recognition remains a research area for at least 50 years and automation, which was used primarily to automatically run scheduled maintenance, updating corporate systems on the same network, but little research was done on using it for improving the usability of a system. In the course of the dissertation, different possible ways of recognizing speech and of automating Windows applications are explored. A prototype is being developed in this research, in which a broad aspect of the platform's capabilities was demonstrated rather than a single aspect in detail. As a part of evaluation, a focus group study was conducted to brainstorm on futuristic scenarios that could benefit using this platform. Finally, the dissertation concludes with advantages and disadvantages of the developed technology.