Speaking Skills Training: A use Case in VET and Adult Learning Environments

Abstract

This paper describes a speech analysis technology that computes personality traits of a speaker. The technology analyses primary speech features via sound engineering and complex personality features (concretely, the “Big 5”) via an AI model trained on a variety of native and non-native speakers of English. The results are communicated to the users/learners so that they can understand how they are being perceived as speakers and therefore improve their speaking skills. To this end, an online course explains speaking features, interpreting personality computing results, and exercises on breathing, etc., to improve one's speaking performance. We describe the process of deployment of this course to learners of VET or other professional education and training at institutions in 8 countries and online globally, including the presentation of the course and the related technology, the identification of participants, and the implementation of a pre-survey on attitudes related to speaking and a feedback survey post-course. We report an analysis of the results of the surveys as well as of the achievements of the learners who took the course. Results suggest that the learners present an improved understanding of how they are being perceived as speakers, as well as a high satisfaction rate and engagement with the course. Additionally, in the short duration of the trial, a slight improvement in speaking skills is arguably already detectable. However, actual improvement in speaking skills is a long-term effort, and generalised claims cannot be made, in contrast with the improvement in understanding how one is perceived by others as a speaker, which is evident by the results presented in this work.