Soundex Algorithms currently used are not working with indian names, Therefore this project will concentrate on generate same soundex for the strings/texts that generates similar sound while pronounced.
Most of the other implementations of soundex finds and removes vowels(a,e,i,o,u,s) in their first step. But what I thought is these vowels or combination of vowels shape the sound of a letter, therefore we must tokenize them first and I tried to do so in my implementation.
All the letters or combination of letters(substrings) replaced or tokenized in this approach are with consideration of Indian Names only. So it will work with indian names in first place.