SPEECH RECOGNITION AND SYNTHESIS AS AN ONLINE LEARNING TOOL: CHALLENGES AND POSSIBILITIES

In recent years, the advent of computerized Speech Synthesis and Speech Recognition has provided a new interface to mobile phones and computer systems through systems such as Apple’s Siri and Google Assistant. Furthermore, the World Wide Web consortium, the main consortium that defines the open standards for the Web, have defined a Web Speech API Specification that standardizes how browser manufacturers and web developers can incorporate speech recognition and synthesis into web sites.

At this time, however, full support for the API is incomplete and variable amongst browser makers, particularly in the mobile space, although it is generally possible to make use of core features in any major OS.

Speech synthesis and recognition has great potential within education where instructors can simply type a dialogue, and have students role play with the computer, which itself can play multiple characters using different voices. Furthermore, through speech recognition student speech can be automatically assessed for accuracy and speed meaning hundreds or even thousands of student assignments can be graded instantly.

However, being a new technology there have been few studies looking at the use of this technology in the classroom. As part of a 3-year cross-institutional research grant from the Japanese Government (Kakenhi), this presentation will look at
(1) the challenges of setting up an online speaking system,
(2) the response of students using the system,
(3) the effect on the student’s performance, and
(4) an analysis of the accuracy of the students’ speech and the system.

The talk will also demonstrate the system in use.