hindi-xlit 1.0.0

A Hindi transliteration library based on AI4Bharat's IndicXlit

Hindi Transliteration Library

A Python library for transliterating English text to Hindi using AI4Bharat's model. This library provides a simple interface for converting English words to their Hindi (Devanagari) equivalents.

Features

• Simple and intuitive API
• Supports both single word and batch transliteration
• Returns multiple transliteration candidates
• CPU-only implementation (no GPU required)
• Interactive command-line interface

Installation

pip install hindi-xlit

Usage

Basic Usage

from hindi_xlit import HindiTransliterator

# Initialize the transliterator
transliterator = HindiTransliterator()

# Transliterate a single word
word = "namaste"
candidates = transliterator.transliterate(word)
print(f"Transliteration candidates for '{word}':")
for i, candidate in enumerate(candidates, 1):
    print(f"{i}. {candidate}")

# Transliterate multiple words
words = ["hello", "world"]
results = transliterator.transliterate_batch(words)
for word, candidates in zip(words, results):
    print(f"\nTransliteration candidates for '{word}':")
    for i, candidate in enumerate(candidates, 1):
        print(f"{i}. {candidate}")

Command Line Interface

The library provides an interactive command-line interface:

hindi-xlit

Available commands:

• Type a word to transliterate
• Type multiple words separated by spaces for batch transliteration
• Type topk N to change the number of candidates (default: 5)
• Type help to show help
• Type quit or exit to exit

API Reference

HindiTransliterator

The main class for Hindi transliteration.

Methods

• transliterate(word: str, topk: int = 5) -> List[str]

  • Transliterates a single word to Hindi
  • Returns a list of top-k transliteration candidates

• transliterate_batch(words: List[str], topk: int = 5) -> List[List[str]]

  • Transliterates multiple words to Hindi
  • Returns a list of lists, where each inner list contains top-k transliteration candidates for the corresponding word

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

This library uses the transliteration model from AI4Bharat.