Week 35: AI is taking over

Project: Whisper AI

Started on:  Feb 19, 2023

Staff Guidance: Chris

Description: This is a very interesting project where I need to dive into the world of AI and text-to-sound capabilities of AI’s model. The main task of this project comprises of exploring different values for a parameter of the Whisper AI model, especially the hallucination silence threshold. This parameters sets out the time limit for when the model can seek past that duration and reprocess at that point after it encounters moments of silence. Therefore, in this project, I need to loop over different parameters and examine the quality of the output in terms of text and the time it takes to output. If at some values of the threshold that we can see improvement in the time and significant changes in the text output, then we can conclude we have found the value that optimizes the model and give us the best result.

 

Step by Step tutorials

• 1) Install Open AI Whisper AI
• 2) Write out a function to extract 10 minutes of the videos
• 3) Write out a function to utilize the Whisper AI model to transcribe the video
• 4) Capture and export the text of the generated caption into a text file