DIGITAL LIBRARY
AI BASED GESTURE AND SPEECH RECOGNITION TOOL FLOW FOR EDUCATORS
1 HTW Berlin - University of Applied Sciences (GERMANY)
2 HU Berlin (GERMANY)
About this paper:
Appears in: ICERI2022 Proceedings
Publication year: 2022
Pages: 6395-6400
ISBN: 978-84-09-45476-1
ISSN: 2340-1095
doi: 10.21125/iceri.2022.1596
Conference name: 15th annual International Conference of Education, Research and Innovation
Dates: 7-9 November, 2022
Location: Seville, Spain
Abstract:
In this paper, we present a tool flow which provides an opportunity for users to transfer their gesture and speech to a robot. Based on our design requirements to protect user's privacy and ensure the system's sustainability and compatibility by engaging users as participants, our goal is to create a user-friendly software system which allows a user with less technical background to animate a robot. As a core part of this system, we created a python-based tool flow which utilizes Mediapipe for motion detection and Vosk for automatic speech recognition. The tool flow, we present in this paper, is conceptually generic enough to be adapted for scenarios using various robots. As a proof of concept, we applied this tool flow to realize a robotic teacher by using an intelligent tutoring system (ITS) and a humanoid robot.
Keywords:
Mediapipe, teacher robot, gesture recognition, automatic speech recognition, tool flow, Vosk.