Transliteration is the process of converting texts from one script to another based on the phonetic similarity. The dictionary meaning says that it is 'writing words or letters in the characters of another alphabet.' This process is only concerned with the pronunciation of the text rather than going into its meaning. Here, the text is displayed in alphabets of different languages but the language, grammar, and sense of original text remain intact in these new characters. For example, the Indian name 'দৃষ্টি’ is pronounced as Dr̥ṣṭi. Its’ transliteration into English is 'Drishti.' We write it as ‘дришти' after its' transliteration into Cyrillic text used in the Russian language.
1.Data :
- Custom dataset : https://github.com/shiv2398/transliteration_english_to_hindi/tree/main/data
- Kaggle dataset : https://www.kaggle.com/datasets/sahityaarya/data-for
- Preprocessed dataset : https://github.com/shiv2398/transliteration_english_to_hindi/blob/main/english_hindi.csv
2.Models :
- Encoder-Decoder Model (Scaler dot Product Attention) :https://github.com/shiv2398/transliteration_english_to_hindi/blob/main/testing_transformer.ipynb
- Encoder-Decoder Model : https://github.com/shiv2398/transliteration_english_to_hindi/blob/main/testing_transformer.ipynb
- Encoder-Decoder Model(Bahandau Attention) : https://github.com/shiv2398/transliteration_english_to_hindi/blob/main/Transformer%20_Models.ipynb