Sub-sequence Matching Algorithm for Improving Automated Speech Recognitions for ORF Assessment

• In Event: Automated Assessment of Writing and Reading Proficiency

Abstract

Commercial speech recognition tools now offer linguistic features that may enable automated methods for evaluating oral reading fluency (ORF). We propose how these features can be leveraged to obtain words correct per minute (evaluated against a human scorer), and discuss modeling procedures to make ORF robust to pronunciations and phrasing.

Authors

• Yihao Wang, Southern Methodist University

• Eric C. Larson, Southern Methodist University

• Akhito Kamata, SOUTHERN METHODIST UNIVERSITY

• Joseph F. T. Nese, UNIVERSITY OF OREGON