VOICING THE UNSAID(VTE)

WHERE WE STAND FOR YOU WHEN NO ONE ELSE

This project deals with the creation of a device that can be used by specially abled or blind people. We already have devices like this but our idea is a bit different, that can directly convert speech to text ,further nobody has to type which saves the clients’ time and is easy to use hence, not as challenging as it used to be.This app has been created using concepts of NLP i.e(Natural processing language). Basically by using NLP and some basic algorithms we aim at creating an app that can simply convert speech into braille and is beneficial in all perspectives.

WHY THIS PROJECT ?

In one of our meetings, we were conversing about influential women and were inspired by the words and thoughts of Haben Girma who is an American disability rights advocate, and the first deafblind graduate of Harvard Law School. So, we decided to take this up as a project.Furthermore, everybody was curious about machine learning.

Our team has different skill sets so accordingly we are designing the project. We all want to deep dive into this field so that it can boost our skills as a team. Our project deals with machine learning algorithms which require a problem-solving attitude so that we can solve real life complex problems easily. We chose this project as it revolves around how we can benefit blind or specially abled. Our project will directly convert speech to braille without any third party and saves time .This project is realistic as we produce something beneficial to the world.

AIM OF THE PROJECT

The aim of the project is that by using speech recognizer (NLP) and converting that text to braille which is useful for the visually impaired. The first parameter that we see in our project is the theory required like acoustic model, and key terms like Automatic Language Recognition, Automatic Speech Recognition (ASR), Bottleneck MLP, CDHMM, Deep Neural Network (DNN) and language models and many more. The next parameter we take into consideration is mastering NLP (natural language processing) concepts, machine learning and its implementation in the project. We are even planning to make a model that can be connected to an app for voice recognition and printing the braille language which would be the driving force and the link of our project.

WHAT PROBLEMS WILL IT SOLVE ?

This project aims to solve real-world problems, making things faster and more helpful. This model lays strong emphasis on the visually impaired and helps them in communicating to people and identifying things quicker. It converts speech to text (for the time being in english) and then converts it to braille hence, there is no need to type as it is time consuming. It will save time and the problem will be solved by our model. This project enables the user with fast and simple layout to understand, in return helping them become independent.

USER INTERFACE DESIGN

We have used Figma as (User Interface Design) tool. The layout contains the Welcome page, Login page, Registration page and Page with Speech to Text, Text translator, History and Terms and Conditions.In the login page, an existing user can directly login with his VTE Credentials (user id and password). Using the registration page, new users can join us using their email id and further VTE Credentials (user id and password ) with a form to fill their personal details would be given using their registered email ids. When a user clicks on speech to text icon a new page will appear where he/she can speak and get the text on the screen and when the user clicks Text translator icon a new page where he could write the text and the written text could be heard.Similarly, terms and conditions can be seen in the terms and conditions icon and history can be viewed and deleted by clicking on history icon.

FLOW DIAGRAM OF OUR PROJECT

WEEK WISE UPDATES FOLLOWED BY SKILLS WE HAVE LEARNED

WEEK 1–2 : Gathering information on basic topics using research papers(extensively) and Internet browsing. Resources for topics like NLP (Natural Language Processing) libraries required for text to speech and speech to text conversion in python followed by App development.

WEEK 3 – 4 : Basic prototypes required for the project and overcame all problems related to libraries using different sources. Design and layouts were made in UI prototype and in android studio followed by tracing constant feedback received from our mentor.

WEEK 5 — 8: Started working on Natural Language Processing advanced implementation, learn and work on the Artificial Intelligence systems, in the process we thought of working on the basic AI automation code (“Prototypes”) .

WEEK 9 — 10 : Building the program and algorithm for the speech to the braille script conversion and started testing the app and overcame any bugs and errors followed by constant feedback from our mentor.

WEEK 11 – 12 : Obtained constant feedbacks and suggestions from industry experts ,mentors and our seniors and modified the project according to it.