Automatic speech recognition technique for voice command

In this paper, we represent a new method for robust speech recognition technique. Speech recognition is a process by which system is able to understand natural language. By the help of proposed method we will be able to give computer system the voice command to perform task. The first step is to take a single audio as an input. The second step is to eliminate the presence of noise. The third step is to match the patterns to recognize the input audio. The error tolerance mechanism is implemented so that it can be robust against human errors, environment noise. There is no requirement of storing the audio for matching, which leads to decrease the space and time complexity both compare to other systems. The database used in the experiment will only be consisting of various voice patterns. Patterns are hard coded in such a way that system is not needed to be trained by the end users.