AI text humanizer designed to make AI-generated content undetectable by services like Turnitin and GPTZero. It targets users needing to mask AI authorship, offering a free, open-source solution that enhances text naturalness and stylistic diversity across multiple languages.

How It Works

The pipeline leverages a multi-stage process: first, DeepSeek rewrites text and translates it to Chinese to introduce structural variation. Second, Google Translate converts this to Turkish, adding syntactic distortion via cross-lingual differences. An optional DeepL translation to Japanese further enhances linguistic diversity. Finally, DeepSeek translates the text back to the original language, aiming to restore readability and natural flow while preserving semantics and evading AI detection. This approach introduces significant linguistic variation to bypass detection algorithms.

Quick Start & Requirements

1. Clone the repository: git clone https://github.com/khrisat/text-humanizer.git
2. Navigate to the directory: cd text-humanizer
3. Install dependencies: pip install -r requirements.txt
4. Copy configuration: copy .config\config.example.toml .config.toml
5. Run the humanizer: python main.py

Prerequisites:

• Python environment.
• DeepSeek API key (required).
• DeepL API key (optional, for enhanced diversity). Configuration is managed via config.toml, allowing specification of target language, API keys, and LLM parameters like temperature (1.3 recommended).

Highlighted Details

• Aims to make AI-generated text undetectable by services like Turnitin and GPTZero.
• Supports 8 languages: English, Japanese, Chinese, Korean, German, French, Spanish.
• Employs a unique multi-stage rewriting and translation pipeline involving DeepSeek, Google Translate, and optionally DeepL.
• Fully free and open-source project.

Maintenance & Community

No specific details on maintainers, community channels (like Discord/Slack), or roadmap were provided in the README.

Licensing & Compatibility

The project is released under the MIT License, which is permissive and generally allows for commercial use and integration into closed-source projects.

Limitations & Caveats

Requires API keys for DeepSeek (essential) and DeepL (optional). Effectiveness against evolving AI detection models is not guaranteed. The multi-stage translation process, while key to its function, may introduce subtle semantic shifts or artifacts.