Open Source ER Design Tools with Voice Input Mode
Oral Defence Date:
Professors Murphy and Wong
The amount of information produced in the world is increasing every day and this rate is expected to continue increasing in the coming years. Data Models are used to describe data for storage and management using Database Systems technology. The ER Design Tools are data modeling tools developed as instructional software at SFSU to illustrate introductory principles for database design. These tools allow ER or UML schema entry and mapping to other data models. Schemas can be displayed in a variety of different modes. For example, the ER Design Tools can take input as an ER schema or XML Schema and generate output as ODMG, Relational or Spreadsheet schemas. Each schema can be displayed in a variety of different modes, such as Data Definition Language (DDL), 2-D graphics, and 3-D graphics. Many students have contributed to the ER Design Tools project as their Cumulative Experience and Master's thesis project under the guidance of Prof. Marguerite Murphy. The new Restructured Implementation of the ER Design Tools is based on Java Web start technology and will support all of the existing functionality and is targeted for Open Source distribution. This report describes the design, logic and the implementation for the voice input mode for the Open Source ER Design Tools. Speech interaction is useful for users who want to reduce the use of keyboard and mouse, and in some cases to increase productivity. An existing Open Source Speech Recognition system (Sphinx4), along with an Speech Synthesis system (FreeTTS) are used to implement the Voice Input mode. The Open Source ER Design Tools with Voice Input mode allows several advantages: minimal use of keyboard and mouse, two way communications between the application and user to build ER Schema and map the application schema to other data model with various output modes. Because the ER Design Tools application is built on the Java platform, we will also discuss speech recognition performance issues on the Java platform. The ER Design Tools with Voice Input mode software is fully functional, and is being released under a GNU3 Open Source license.
Speech Recognition, Speech Synthesizer, Grammars, Decoding, Mapping