bhāṣā śālā
language house
An etymologically-informed reader for Indian language texts. Word-level transliteration, translation, and etymological connections across hundreds of Indic languages.
Every word gets transliteration, translation, grammar, and morphology. Hover any word to see its full breakdown with provenance from multiple dictionaries.
See where each word comes from — trace its journey from Proto-Indo-European through Sanskrit, Prakrit, and into modern Indic languages with interactive trees and maps.
Cross-referenced dictionary compiled from Wiktionary (17 languages), Jambu/CDIAL (290K forms), Monier-Williams (286K entries), and Heritage Engine (928K Sanskrit forms).
When a word has multiple meanings, an LLM (Llama 3.3 70B) reads the surrounding context to select the right sense — then caches it so no API call is ever repeated.
See where a word's relatives live — cognate forms across 615 languages plotted on an interactive map, powered by Jambu etymological data.
Every meaning and etymology shows its source — dictionary badge, confidence score, and direct links. AI-generated glosses are clearly marked and cross-checked.
PDFs are processed with Gemini 2.5 Flash to extract text and word bounding boxes
Each word is looked up across Wiktionary (16 languages), Jambu/CDIAL (290K forms), and Monier-Williams (160K entries)
An LLM (Llama 3.3 70B via Groq) selects the best sense for each word in context
Cross-linguistic cognate data spanning hundreds of Indic languages is assembled into interactive etymology visualizations