DyslexiaElevate

Spelling Exercise Interface
Image-to-text-to-speech Converter
Reading Exercise Feedback
Analyzing spoken words in reading exercise
Home Interface
About Us
Features

Inspiration

We have witnessed firsthand the challenges that some children face when it comes to reading. Their struggles inspire us to create DyslexiaElevate. We are dedicated to creating a digital platform that embraces diversity, breaks down barriers, and empowers individuals of all backgrounds and abilities to access, understand, and engage with information seamlessly. We believe that education is a powerful tool for empowerment. Through our features like spelling and reading exercises, we strive to enhance language skills for dyslexic learners.

What it does

Our web platform is dyslexia-user friendly as all fonts on it are dyslexia friendly. Users can also upload files like pdf or docs with texts to our dyslexia-friendly font converter, this allows the texts in the file to be written in a dyslexia-friendly font(with the exact format!). Further, rather than allowing users to listen to all the texts at once, we allow users to click on the specific words they want to listen, less time-consuming for users if they already know the other words. This distinguish our platform from audiobook!

Dyslexic learners also have difficulties in reading and spelling, and that is why we make those exercises in our platform. In the reading exercises, users are asked to record themselves reading a one-sentence story, they will get a score and feedback as well. Users can then know the specific words they got wrong, and they can repeat practicing that word before moving on to another story. To make the learning fun, users will get AI-generated drawing of the story they previously have completed correctly. Note that users can replay the word they got wrong so they can know how they sounded. (All the stories are generated by AI)

For the spelling exercises, users get to copy the words as how they see it. If they got it wrong, they have to repeat until they copy it accurately. To make the feature user friendly, we also allow them to listen to the word by clicking the word itself and whenever they got the spelling wrong, the part where they got the spelling wrong would be highlighted in red. When users are done practicing, they will get cute pictures (chibi), the chibi will be different if they get different score. This gives the learners some sort of achievement! (All the words are generated by AI)

Three main features of our web platform:

Image-to-text-to-speech converter
Reading Exercise:
Spelling Exercise

How we built it

We built the frontend using html/css and js. Using flask as our main framework. Utilizing pytesseract OCR model for our Image-to-text-to-speech converter, google-cloud API for reading and recognizing words, and cohere for generating story for our reading exercise.

Challenges we ran into

We faced difficulty when implementing an algorithm that can format converted text so that they follow the same pattern as the original text. We also faced challenges when debugging our code because we didn’t make any unit test which was a really big problem. For the interface, we had different devices with different aspect ratios, and the CSS code hat works great on the first ratio, give us an unexpected looking website on device with another ratio. We just learned CSS three days ago, so this website is our first try!

Accomplishments that we're proud of

We managed to create a fully functional website for the first time for both of us! Further, we managed to include many useful additional features beyond our initial plan, with enough time for debugging.

What we learned

Back-end -> We learned how to use several machine learning API. Front-end -> We learned how to design a good-looking website using CSS and responsive navigation bar