https://github.com/carlosdavidantas/luminai
AI-powered tool that transcribes YouTube videos and answers your questions with contextual insights, summaries, and explanations.
https://github.com/carlosdavidantas/luminai
ai chatbot chromadb huggingface langchain rag transcription whisper
Last synced: 3 months ago
JSON representation
AI-powered tool that transcribes YouTube videos and answers your questions with contextual insights, summaries, and explanations.
- Host: GitHub
- URL: https://github.com/carlosdavidantas/luminai
- Owner: carlosdavidantas
- License: mit
- Created: 2025-02-12T21:28:15.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-04-11T23:02:35.000Z (about 1 year ago)
- Last Synced: 2025-04-11T23:32:39.593Z (about 1 year ago)
- Topics: ai, chatbot, chromadb, huggingface, langchain, rag, transcription, whisper
- Language: JavaScript
- Homepage:
- Size: 171 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Luminai
AI-powered tool that transcribes YouTube videos and answers your questions based on their content.
## About
Luminai was created to make consuming YouTube videos easier and faster.
You can ask Luminai to simplify and summarize the key points of a video, for example, a long podcast, helping you quickly grasp the most important information.
Powered by AI, Luminai has context about the selected video and memory of your conversation, so it remembers what you’ve discussed.
The workflow is simple:
1- Luminai downloads the audio track of the chosen video.
2- The audio is transcribed into text.
3- This text is converted into vector embeddings and stored in a database.
4- The stored data is then used as context for the AI to answer your questions.
## Installation & Setup
#### Pre-required:
- **[Node.js](https://nodejs.org/)** and npm
- **[Python](https://www.python.org/)** and pip
- **[FFMPEG](https://www.ffmpeg.org/)** for audio
### Backend Setup (API)
```bash
# Clone the repository
git clone https://github.com/carlosdavidantas/luminai.git
# Navigate to the API folder
cd src/api
# Create a virtual environment
python -m venv .venv # Windows
python3 -m venv .venv # Linux
# Activate the virtual environment
.\.venv\Scripts\activate # Windows
source .venv/bin/activate # Linux
# Install dependencies
pip install -r requirements.txt # Windows
pip3 install -r requirements.txt # Linux
# Create a .env file
HUGGINGFACE_API_KEY=your-key-here
GOOGLE_API_KEY=your-key-here
LANGCHAIN_API_KEY=your-key-here
# Run the API
python api.py # Windows
python3 api.py # Linux
```
### Frontend Setup
```bash
# Open a new terminal and navigate to the frontend
cd src/web-interface/app
# Install dependencies
npm install
# Start the development server
npm start
```
Your default browser will open Luminai at http://localhost:3000.
## Usage Guide
Home screen
Click the “+” button, then click **“Add YouTube Link”**:
Enter your YouTube link in the modal and click **Save**:
Wait while Luminai processes the video:
When it’s ready, your chat will appear on the left side:
Now you can **ask anything** related to the video:
Luminai has **memory**, so it remembers your previous messages:
To start a new chat, click **“New Chat”**, the screen will reset and you can add a new video:
To **delete a chat**, hover over it and click the **delete button** that appears:
You can also **toggle the sidebar** to show or hide your chat history:
## Tecnologies
- React.js
- Javascript
- Flask
- Python
- Langchain
- Retrieval-Augmented Generation (RAG)
- Chroma
- Whisper AI
- Geminai AI
## License
This project is licensed under the **MIT License**.
See the **[LICENSE](./LICENSE)** file for details.
## Author
**[Carlos Davi Dantas](https://github.com/carlosdavidantas)**