Gpt_Builder is a Python application that harnesses the power of GPT (Generative Pre-trained Transformer) to build customized applications based on user descriptions provided through audio recordings and drawings. This README.md file serves as a guide for users and developers interested in understanding, using, or contributing to the Gpt_Builder project.
-
User Input:
- The user starts by recording an audio description and creating a drawing that outlines the desired application.
-
Audio to Text Conversion:
- Utilizing the Whisper Text-To-Speech model, the audio is converted into text.
-
Requirement Extraction:
- The text is processed with GPT-3.5 to create a list of functional requirements using the format: REQ_ID, TYPE, DESCRIPTION.
-
UI Sketch Analysis:
- The drawn image is analyzed with GPT-4-Vision to generate a list of UI requirements using the format: WIDGET_ID, TYPE, LAYOUT, DESCRIPTION.
-
Code Generation:
- Both sets of requirements are combined to generate the code for the requested application with GPT-4.