Inspiration

In the UK, less than half of those from a Bangladeshi background in the UK speak English as their first language. 1 in 8 from other white backgrounds cannot speak English well, or at all. Not being able to speak English properly can limit opportunities for yourself, and can affect the language proficiency of young children being brought up. English is notoriously known as the most difficult language to learn. read.io aims to make the learning of English more efficient, personalised and fun.

What it does

Whether English is your second language or for your child learning to read, read.io makes the learning process a whole lot easier. Our application provides initial reading material from one of our pre-downloaded books, and by using the user's progress alongside Natural Language Processing technology, creates new sentences in the style of the book being read - in particular, including words that have been mispronounced by the user. Through further practise and reinforcement of words mispronounced by the user, learning becomes more efficient and accessible.

How we built it

Being a web-based application, we have used HTML, JavaScript and CSS bootstrap for the webpages. Natural Language Processing has been implemented using nlp.js and Azure Speech-to-Text has been implemented in JavaScript. Pre-hardcoded reading material was sourced from online.

Challenges we ran into

With so many sub-challenges, brainstorming ideas took a while and from climate-based to educational ideas, we were lost for direction! We worked as hard as we could with the time that we had but in hindsight we should settle for an idea early.

Microsoft Azure was unfamiliar technology at first. Hours were spent into implementing the Speech-to-Text in Python. Unable to find a solution and pushed for time, we adapted and decided to implement in JavaScript.

Accomplishments that we're proud of

We are really proud with the progress we have made with Natural Language Processing, since this was unfamiliar technology. It holds a lot of potential and there is a lot we can do with this in future. We are also really proud that we have created something that enables accessibly education - a fundamental human right.

What we learned

This was the first time we have played around with Natural Language Processing technology. After doing much research we determined how best to use and implement it for the greater educational good. In addition, we have previously used Google Cloud SDK text-to-speech, but this was the first time we used Microsoft Azure.

What's next for read.io: Reading, styled to you

Because of the lack of reading material pre-installed, the sentences created using Natural Language Processing aren't always of the most accurate form. In future, we will add more reading material for NLP technology to work with, in turn producing better structured sentences. In addition more books will be added to provide a breadth of reading material for learners, including of different genres and complexities.

Built With

Share this project:

Updates