Automatic assessment system of speaking modules for IELTS

• Main Author: Ho, Sy Dong.
• Other Authors: Hui Siu Cheung
• Format: Final Year Project
• Language: English
• Published: 2010
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: http://hdl.handle.net/10356/42391
• Institution: Nanyang Technological University

English is an international language which is used popularly over the world today. Some of English tests such as TOEFL or IELTS become a standard for evaluating English proficiency of individuals in academic settings. In general, an English test consists of 4 modules: listening, reading, writing and speaking. However, the assessments of these tests are greatly time-consuming and need a lot of effort from assessors, especially the speaking module. This project aims to develop a web-based application that can automatically evaluate online speaking English tests. In the scope of this project, testing data follow the format of IELTS speaking module. For the purpose of evaluating oral tests, some techniques of speech recognition and text-tospeech are studied and compared with regard to their performances. Voices of users are live streamed to server for processing through Real Time Messaging Protocol (RTMP). There are two parts of the evaluation: grammar and semantics. The rule-base grammar checking technique researched by Stanford University is employed. Grammar of each sentence and style of the overall speaking or essay are assessed by this technique. On the other hand, for semantic checking, experiments of ensemble of semantic analysis techniques are conducted to compare with other techniques. This report provides detail information of the structure of the project and also gives recommendations to improve the performance of current application.