Inspiration
As a child grows up, developing their reading aloud skills is very important. If they have trouble with a word, someone should be there for them to tell them how to say the word. We want to lessen a child's speech impediments.
What it does
We ask for a file for the book, you read a paragraph at a time, and we will tell you what words you struggled with (and how to properly say them).
How I built it
We used a Python backend and the IBM Watson speech-to-text API. It will develop a string for us to compare with the original. We look for hesitations and output the words around the hesitation
Challenges I ran into
It was hard to quantify what an "impediment" was. Due to a limited time, we picked one (hesitations) and focused our efforts to output what we needed. Front end is hard.
Accomplishments that I'm proud of
We went over a lot of ways to detect speech impediments and thought it through in a theoretical way.
What I learned
2 back end programmers should have a front end programmer on their team.
What's next for Reader Bear
We will make the product more child-friendly with improved UI/UX, and maybe on a different medium (app). It will output diagnostics for the child and the parent to see improvement over time. We will use machine learning and train our assistant to detect speech impediments based on a particular child's speech patterns. It will train the data with increasing amounts of voice recordings. We will cover more types of speech impediments. (kachow)
Log in or sign up for Devpost to join the conversation.