alakxender/dhivehi-translit-mixed

Dhivehi Transliterated Mixed Dataset

Dataset Summary

This dataset contains Dhivehi sentences and titles transliterated into Romanized Latin script. It is intended to support tasks like transliteration, normalization, pronunciation modeling, and text-to-speech for Dhivehi.

The data includes a mixed set of:

• Complete Dhivehi sentences
• Titles of articles and phrases

Languages

• Source Language: Dhivehi (Thaana script)
• Target Language: Dhivehi (Romanized Latin)

Dataset Structure

• original: The original Dhivehi text in Thaana script
• latin: The transliterated text in Romanized Latin

Example

{
  "original": "މިއަދު ހިނގި ވަރަށް",
  "latin": "miadhu hingi varah"
}

Use Cases

• Training transliteration models
• Preprocessing for text-to-speech (TTS)
• Making Dhivehi text easier to search or type in Latin script
• Building tools for Dhivehi in low-resource NLP

Dataset Creation

This dataset was automatically generated and not manually refined. It may not reflect the most accurate or natural transliterations.