The aim of this paper is to present Vowel recognition for Hindi language. The vowel recognition is an important step for developing speech recognition system. Thus there is a need to explore vowel recognition related issues for obtaining better speech recognition results. Experiments were conducted using Connected word Hindi speech corpus for Speaker dependent mode using widely used Hidden Markov Model(HMM) based HTK Tool kit for both training and testing. Hindi Speech Corpus, made of 600 utterances spoken by 5 speakers, was used in this experiment. Mel Frequency Cepstral Coefficients (MFCCs) were used with 5 states monophone based HMM model for feature extraction. Different Hindi speech characteristics were explored using formant analysis. Experimental results achieved as average vowel recognition scores of 77.12% for front vowels, 84.4% for middle vowels and 86% back vowels. Average vowel recognition score was achieved was 83.19%. Finally paper concludes future development direction.