---
title: NewsIA
emoji: 📰
colorFrom: blue
colorTo: red
sdk: gradio
sdk_version: 5.0.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: 'AI-Powered News Generator'
---

# 📰 NewsIA - AI News Generator

NewsIA is an application that uses artificial intelligence to generate professional news articles from multiple sources such as documents, audio, video, URLs, and social media.

## Features

- **Multiple input sources**: Documents (PDF, DOCX, XLSX, CSV), audio, video, URLs, and social media.
- **Automatic transcription**: Automatically transcribes audio and video files.
- **Information extraction**: Extracts relevant content from web pages and documents.
- **Optimized generation**: Uses optimized models to generate news quickly and efficiently.
- **Customization**: Allows adjusting the tone and length of generated news articles.

## Technologies used

- **TinyLlama**: Base model optimized for performance in environments with limited resources
- **Whisper**: For audio and video transcription
- **Gradio**: For the user interface
- **Hugging Face Spaces**: For hosting with ZeroGPU

## How to use

1. Provide **instructions** and **main facts** for your news article
2. Adjust the desired **length** and **tone**
3. Add additional sources (optional):
   - **Documents**: Upload PDF, DOCX, XLSX or CSV files
   - **Audio/Video**: Upload audio or video files for transcription
   - **URLs**: Add links to relevant web pages
   - **Social Media**: Add links to social media posts
4. Click on **Generate News**

## Installation

To run this application locally:

1. Clone this repository
2. Install the required dependencies with `pip install -r requirements.txt`
3. Run the application with `python app.py`

## Environment Variables

The application uses the following environment variables:

- `HUGGINGFACE_TOKEN`: Your Hugging Face API token (required to access the models)

## Technical Details

NewsIA uses a two-stage approach:
1. **Data Processing**: Extracts and processes information from various sources
2. **News Generation**: Uses a language model to create a coherent news article based on the processed data

The application is designed to run efficiently on Hugging Face Spaces with ZeroGPU resources, using optimized memory management techniques to improve performance.