Context Dependent Speech Recognition Using VITERBI Algorithm

This research is based on the conversion of English speech into English text with an efficient system for independent speaker speech recognition based on Neural Network Approach using Viterbi Algorithm. To recognize the English words consider all the accents of same spoken word, so that matching process with the actual word does not lead to any difficulties. There are 26 characters in English. It is well known that the pronunciation of a word depends heavily on the background. Speech dependent & speaker independent technique can be used and English words must be recognized.
HMM n to 1 encoder and HMM 1 to n decoder for finding the speech to text with the help of viterbi algorithm. There may be two or more pronunciation for the same word so the database should be maintained in that all expected accent should be there for a particular word. The main focus is to clearly get the spoken word for the different pronunciation/accent. This approach is better than others as the identification becomes easier.