Document Type
Honors Thesis
Abstract
Methods of knowledge transfer that rely primarily on visual and/or auditory formats do not effectively convey context-specific or implicit skills, known as tacit skills. This limits knowledge transfer. In this work, the use of customizable pitch captions and spatial audio vibration captions is proposed to aid in conveying this tacit knowledge for neon glass bending video tutorials. Such a system is designed to provide users with greater control and support, which may maximize the information they obtain from, improve the autonomy they have with, and experience they have with a learning tool. As such, a system interface was developed that allows users to develop customized spatial audio and pitch caption cues to communicate tacit information. This study examines how these cues can be designed by analyzing natural language queries users might ask when watching expert bending tutorials and translating them into audio cues using If-This-Then-That (IFTTT) rules. GPT-3.5 was used to translate three natural language questions about a fictional user’s and two expert benders’ performance into IFTTT rules, generating spatial audio vibration cues. Results indicate that LLMs can effectively transform natural language prompts into IFTTT rules, which may allow for integration of automated cues generated from natural language prompts with video tutorial systems.
Disciplines
Artificial Intelligence and Robotics | Computer Sciences | Glass Arts | Graphics and Human Computer Interfaces
Publication Date
5-2025
Language
English
Faculty Mentor of Honors Project
Cesar Torres
License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.
Recommended Citation
Kapoor, Gunnika, "CueGen: Customizing Sensor Captions For Neon Bending Tutorials" (2025). 2025 Spring Honors Capstone Projects. 21.
https://mavmatrix.uta.edu/honors_spring2025/21
Included in
Artificial Intelligence and Robotics Commons, Glass Arts Commons, Graphics and Human Computer Interfaces Commons