The Speech recognition systems are difficult and challenging as the same word can be spelled in many ways also change in accent has a huge impact on the accuracy of speech recognition. Most of the ASR systems in use today are designed to recognize speech in English. The major objective of this research is to design an ASR system which recognizes discrete words from Hindi language & controls some action depending on the provided input. The input voce is captured using a microphone which are then preprocessed using several algorithms like Dynamic Time Wrapping (DTW), Hidden Markov Model (HMM) etc. This paper aims at developing a simplified technique for recognition of speech spoken in the Hindi language by first modeling the system on computer-based design and then deploying it on an embedded system.