An open-source speaking practice and testing application
Recent advances in AI and neural machine learning have significantly improved the accuracy of automatic speech recognition (ASR) technologies. Consequently, ASR tools have been gaining headway in language learning environments. The presenter will first introduce recent trends in speech recognition and speech assessment and review a number of auto-graded speech tools that are currently available to teachers. The main portion of the presentation will focus on an open-source speaking practice and testing application that is being developed by the presenter. This web-based application enables teachers to easily create customized speaking tasks with embedded text, images, audio or video prompts. The custom speaking tasks are automatically scored and immediate feedback is provided at the individual level. The application integrates Google’s speech recognition engine with a phoneme-based text comparison algorithm to automatically generate a speaking score. A number of speaking activities that can be deployed and scored using this application will be demonstrated. Examples of imitative, intensive, responsive, interactive, and extensive speaking tasks will be provided. The software is available as open-source code on GitHub, and is compatible with the latest versions of the Moodle course management system.