Skip to content
View Blessy7-eng's full-sized avatar

Block or report Blessy7-eng

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Blessy7-eng/README.md

🎬 PulsePoint AI: Viral Reel Engine

PulsePoint AI is an automated video-to-reel engine built for the ByteSize Sage Hackathon. It transforms long-form educational content into high-impact, 9:16 vertical reels using local audio analysis and speech-to-text.

🚀 The Core Problem

During the hackathon, we found that relying on cloud-based LLMs (like Gemini) created a massive bottleneck due to strict rate limits and 429 RESOURCE_EXHAUSTED errors. We needed a solution that was fast, free, and worked 100% offline.

✨ Our Solution: The Local Pulse Engine

PulsePoint AI bypasses cloud limitations by using a Hybrid Local Processing approach:

  • Audio Energy Peak Detection: Uses librosa to analyze the mathematical "Pulse" of the audio. It detects loudness spikes (RMS analysis) to automatically find where the speaker is most engaging.
  • Local Transcription: Runs OpenAI's Whisper (Tiny) model locally to generate timestamped transcripts without any API keys.
  • Smart 9:16 Cropping: Automatically centers and crops 16:9 landscape video into mobile-ready vertical reels using MoviePy.
  • Dynamic Captions: Synchronizes Whisper’s timestamps to overlay bold, social-media-style captions directly onto the video.

🛠️ Tech Stack

  • Frontend: Streamlit
  • Audio Analysis: Librosa & NumPy
  • Transcription: OpenAI Whisper (Local)
  • Video Engine: MoviePy
  • Environment: Python 3.9+

⚙️ Installation

Prerequisites

  1. ImageMagick: Required for rendering text captions.
    • Windows: Install ImageMagick and update the IMAGEMAGICK_BINARY path in main.py.

Setup

  1. Clone the Repo: 
    git clone [https://github.com/Blessy7-eng/ByteSize_Sage-Hackathon.git](https://github.com/Blessy7-eng/ByteSize_Sage-Hackathon.git)
cd ByteSize_Sage-Hackathon
  2. Install Dependencies: 
    pip install -r requirements.txt
  3. Run the App: 
    streamlit run main.py

App Working video : https://drive.google.com/file/d/1LNqH5f7-TEmuNp5KUZCja4fLVVOCBUNv/view?usp=sharing

reel 1 : https://drive.google.com/file/d/1uqT_qrceuJb6LR6oUxgmzl8C_tCNkBUq/view?usp=sharing reel 2 : https://drive.google.com/file/d/1uqT_qrceuJb6LR6oUxgmzl8C_tCNkBUq/view?usp=sharing reel 3 : https://drive.google.com/file/d/1iSHfnAJDrtbPzFyoOV0fHLwjRvFyCOAg/view?usp=sharing reel 4 : https://drive.google.com/file/d/1c4ZENq3olUbAqoguqxJ9UBgYLoZnNDgx/view?usp=sharing

Popular repositories Loading

  1. Blessy7-eng Blessy7-eng Public

    Config files for my GitHub profile.

    Python

  2. Spotify-Clone- Spotify-Clone- Public

    This is the Spotify clone created using HTML, CSS, JAVASRIPT. I included copyright as well as non-copyright songs in this website just as an clone website not for personal use.

    HTML

  3. Netflix-Clone Netflix-Clone Public

    This is a Netflix Frontend Clone.

    CSS

  4. Fredoc-App Fredoc-App Public

    It is a Health Awareness/ Disease Awareness App, Named 'FreDoc' combined with a name 'Friend' & 'Doctor'. The app also have a chatbot, which will guide related to our health.

    Dart

  5. dubbo dubbo Public

    Forked from apache/dubbo

    The java implementation of Apache Dubbo. An RPC and microservice framework.

    Java

  6. dubbo-website dubbo-website Public

    Forked from apache/dubbo-website

    Apache Dubbo documents

    CSS