Brainrot to AI

Input Screen
Video output
Complete App

Inspiration

My inspiration for this project came from the viral "brainrot" videos on YouTube, TikTok, and Reels. These typically feature an AI voice reading a script (often a Reddit story) with captions on the screen, paired with Minecraft gameplay or other goofy background footage.

What it does

My app generates these types of brainrot videos in minutes, minimizing the time and human input needed to create this content. Here’s how it works:

Video Script Input:
- Type or paste your own script.
- Generate a script using Groq AI after choosing a story genre.
- Enter a Reddit subreddit to fetch a trending story for your video script.
Brainrot Transformation: The app allows you to convert your script into "Brainrot" (optional). This modifies the script by replacing common terms with popular "brainrot" phrases like Skibidi and Rizz.
AI Voice Selection: Choose the AI voice that will read your script from a dropdown menu below the textbox.
Generate Video: Press the "Generate Video" button to process and create the video. The steps include:
- Audio Generation: Using Edge-TTS to generate the AI voice audio.
- Subtitles Creation: Using Whisper API, FuzzyWuzzy, and a custom algorithm to improve subtitle accuracy.
- Background Video Clipping: A random start time is chosen from a background video, and the audio and captions are synchronized with the video, formatted for 9:16 short-form content.
- Final Output: The processed video is delivered to the user for download.

Challenges We Ran Into

One of the major challenges was generating accurate subtitles for the "brainrot" scripts, which often use slang and unconventional words. The Whisper API struggled to correctly transcribe these terms, resulting in inaccurate captions. To address this, I implemented a two-factor system:

Compare the words in the video script with the generated captions to correct any spelling errors.
Use the FuzzyWuzzy library to check for better matches between the script and captions, adjusting the index to realign them in case of desynchronization. Used a buffer region of a few surrounding words to accomplish this.

This approach helps maintain subtitle accuracy, ensuring the captions are as aligned as possible, even if minor errors occur.

Accomplishments We're Proud Of

I’m proud of being able to bring my vision to life by implementing so many ideas into a single, cohesive app. Each challenge I faced while developing this app pushed me to come up with creative solutions, and I’m thrilled that everything came together successfully. I feel like this is an awesome project considering this is my second-ever Hackathon.

What We Learned

This project gave me valuable experience working with various REST APIs and frameworks. It was also the first app I developed where I worked on the frontend. I gained hands-on experience with Flask, HTML, CSS, and JavaScript, frameworks and technologies I had limited experience with before.

What's Next for Brainrot AI

I plan to expand the customization options for the generated videos beyond the script and AI voice, such as offering different background options, user-uploaded videos, and more. I’m also looking forward to hosting the app for others to use and experiment with.