ChatGPT and other generative language models have the potential to revolutionize human productivity and communication. We apply generative language models to PowerPoint slideshows to automate away its infamous tedium.
PrezPal generates PowerPoint slides in real-time during the presentation delivery in response to voice commands. PrezPal leverages generative language models and MATLAB machine learning functions to ensure the constructed slides have comprehensive and concise content, complete with engaging figures.
We made full use of MATLAB functionality, utilizing tools from the Audio, Deep Leaning, and Report Generation Toolboxes. We also learned how to make calls to Open AI API from MATLAB to access GPT 3.5 Turbo.
We want to have live data analysis. As part of the voice command, the presenter should be able to request PrezPal to perform small data analysis tasks on some given dataset and show the resulting figures in the next slide. The code would be written by generative language models. We want PrezPal to be able to anticipate slides based on presentation context instead of explicit voice commands for an even smoother experience. We want to engage with PowerPoint features other than images and text, like videos and animation.
Devpost: https://devpost.com/software/prezpal YouTube: https://www.youtube.com/watch?v=nV2BFURIWzI