YouTube summarizer (Whisper + AI)

A simple tool that takes a YouTube video, extracts the audio, transcribes it, and generates a short summary of the content.

The goal is to quickly understand long videos without watching them fully.

---

What it does

• takes a YouTube link as input
• downloads audio from the video
• converts speech to text using Whisper
• generates a summary using an NLP model (BART)

---

Requirements

• Python 3.10+
• FFmpeg installed and added to PATH
• Internet connection (for downloading models the first time)

---

Installation

1. Clone the repo

git clone https://github.com/RSlucas/video-summarizer.git
cd video-summarizer

2. Create virtual environment

python -m venv venv
Activate it:
venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Install FFmpeg

Required for audio processing.

Download here: https://ffmpeg.org/download.html

Add it to PATH and check:

ffmpeg -version
Usage

Run the program:

python main.py

Then paste a YouTube link:

https://youtu.be/xxxx

Wait for the transcript and summary output.

How it works
yt-dlp downloads audio from YouTube
Whisper converts speech into text
a transformer model summarizes the transcript
Project structure
video-summarizer/
│
├── main.py
├── downloader.py
├── transcribe.py
├── summarize.py
├── requirements.txt
└── .gitignore
Known issues
 -- First run is slow because models are downloaded
 -- Runs on CPU by default (slower but stable)
 -- FFmpeg must be installed correctly or audio extraction will fail

How to use Streamlit

run on cmd: streamlit run app.py

Future improvements

Web UI (Streamlit or React frontend)
Better summarization using LLM APIs or larger models
Export results to PDF or Markdown
Highlight key points instead of full summary

Notes

This is a learning project to explore:

• speech-to-text systems
• NLP summarization models
• YouTube automation pipelines