Human beings developed the ability to communicate with one another at the dawn of civilization and the primary method and approach of that communication methodology was through speech. Language development and the ability to communicate in a common language is the backbone of human living structure. Unfortunately not everyone in this world has the ability to use to speech or conventional hearing methods in order to communicate. They are known as hearing impaired people and they make use of sign language in order to communicate with each other and the rest. According to recent survey by Center of bibliographical study and research in California, 466 million people have hearing disability and 34 million of them are children. Hearing disability creates an issue for them from workplace to household and especially for children in emergency circumstances. The solution that currently exists in order to deal with this problem is human resource based where someone with hearing disability should be able to hire a sign language interpreter in order to communicate with those who do not understand sign language. The issue that comes along with this approach is that this is not very feasible or practical approach for may deaf and mute people and is unable to solve communication problems specially during emergency crisis period. The digital solutions that exist in dealing with this problem have two hindrances that comes along with them: the solution is either very expensive where it goes above $10,000 or the solution is not portable where it could not be carried to a classroom or a workplace. The solution that we have provided uses neural networks to translate sign language gestures into corresponding words, sentences, alphabets and numbers in English Language. The solution is affordable and portable as it comprises of a Raspberry pi, a camera, a potable battery and an LCD screen where the user just has to perform sign language gestures in front of the screen and corresponding results would appear written on the LCD. The way forward with this project is increasing the dataset and diversifying training images in order to increase and accuracy and completely omits the background or light visibility issue. Furthermore, the next approach for the this project could be releasing this project in the form of an application once proper accuracy levels are met in order to make this project completely free and perfectly portable