CoolFace
Apppublic

Ali578/ai-text-humanizer

sourceHugging Facemitupdated 11mo agoView on Hugging Face
2likes
App README

๐Ÿง”๐Ÿปโ€โ™‚๏ธ AI Text Humanizer ๐Ÿค–

Transform AI-generated text into natural, human-like, academic-style writing with advanced NLP techniques.

โœจ Features

  • โ€”๐Ÿ“ Contraction Expansion - Converts informal contractions to formal language (don't โ†’ do not)
  • โ€”๐ŸŽ“ Academic Transitions - Adds professional transition words for better flow
  • โ€”๐Ÿ”„ Passive Voice Conversion - Optional transformation to passive voice for formal tone
  • โ€”๐Ÿ“š Intelligent Synonym Replacement - Context-aware word substitution using semantic similarity
  • โ€”๐Ÿ’พ Export Functionality - Download transformed text as .txt files
  • โ€”๐ŸŽฒ Reproducible Results - Set random seed for consistent transformations
  • โ€”๐Ÿ“Š Real-time Statistics - Track word and sentence count changes

๐ŸŽฏ How to Use

  1. 1.Enter or upload your text - Paste your AI-generated text or upload a .txt file
  2. 2.Select transformation options in the sidebar:
  3. 3.Enable Passive Voice Transformation
  4. 4.Enable Synonym Replacement
  5. 5.Enable Reproducible Results (set seed)
  6. 6.Click "Transform to Academic Style"
  7. 7.Review the results - Compare statistics and download if needed

๐Ÿ› ๏ธ Tech Stack

  • โ€”Framework: Streamlit 1.42.0
  • โ€”NLP Processing: spaCy 3.8.4
  • โ€”Semantic Analysis: Sentence Transformers 3.4.1
  • โ€”Text Processing: NLTK 3.9.1
  • โ€”ML Backend: PyTorch 2.6.0

๐Ÿ“Š Performance

  • โ€”โšก Model caching for 70% faster repeat transformations
  • โ€”๐ŸŽฏ Optimized dependencies (18 essential packages)
  • โ€”๐Ÿ›ก๏ธ Comprehensive error handling
  • โ€”๐Ÿ’พ Minimal memory footprint (~1GB)

๐Ÿ”ง How It Works

  1. 1.Contraction Expansion - Uses NLTK tokenization with smart spacing
  2. 2.Academic Transitions - Adds formal connectors (Moreover, Additionally, etc.)
  3. 3.Passive Voice - spaCy dependency parsing identifies Subject-Verb-Object patterns
  4. 4.Synonym Replacement - WordNet synonyms filtered by semantic similarity (>0.5 threshold)

๐Ÿ“ˆ Use Cases

Perfect for:

  • โ€”Students formalizing casual writing
  • โ€”Content creators humanizing AI-generated text
  • โ€”Researchers improving draft readability
  • โ€”Anyone needing quick academic tone conversion

๐ŸŒŸ Example

Input:

I don't think this'll work. It's really bad.

Output:

Moreover, I do not think this will work. Furthermore, it is really bad.

๐Ÿ“ License

MIT License - Feel free to use and modify!

๐Ÿ”— Links

  • โ€”GitHub Repository: https://github.com/AliMehdi512/AI-Text-Humanizer-App
  • โ€”Documentation: Check the README.md in the repository

๐Ÿ™ Acknowledgments

Built with โค๏ธ using:


Made with โค๏ธ for the AI/NLP community