Humans are considered special compared to other animals because of their ability to speak. Speech allows us to communicate our thoughts to others in a simple form. The world is now such that humans need to interact more with machines. So, if machines could understand our speech, that would be useful. It is therefore essential to allow computers to understand which user is talking about for better human-computer interaction (HCI). Speech consists of audio and visual modalities that provide a certain level of redundancy and exhibit different immunity to noise, which is inevitable in most real-world speech recognition scenarios. Speech recognition using audio signals can be performed in controlled environments. The results are promising when the system operates on a clean audio signal. The presence of noise in the input signal prevents the audio system from accurately identifying speech. Reducing noise using filters will remove noise, but remove some useful signals in the process. Thus, using visual information about the speaker's lip movements as an additional modality will improve speech recognition performance....