Local AI text forensics with llama.cpp using "Binoculars" approach to compute observer perplexity, cross-perplexity, and custom score, with CLI scoring and heatmap generation, and both a VS Code extension and tkinter client.

TextAuth — Evidence-Based Text Authenticity Analysis: A forensic analysis system that evaluates textual evidence using multiple statistical, linguistic, and semantic signals to assess content authenticity across education, publishing, hiring, and research domains.

Trigram Language Model for Spanish trained on Cervantes' texts