The major inspiration was the mutual upbringing of having immigrant parents and them having a difficult time learning English. Additionally, natural language processing was a strong interest shared amongst our group.
What it does
We first allow the user to choose a book and then present slides consisting of small sentences from the book. The user is then required to verbally read each slide and will be informed of their reading accuracy at the end of a given number of slides.
How we built it
Everything was built mainly using the React framework. We used Mozilla's speech recognition model to convert voice to text which allowed us to assess reading accuracy. Finally, we deployed our website using domain.com.
Challenges we ran into
Voice recording was one of the biggest challenges we faced when developing this project mostly because there were many complicated libraries. Additionally, deploying a pre-trained model as a function and a lambda were difficult because these models are trained on relatively large datasets and integrating the model with Functions and Serverless was very difficult.
Accomplishments that we're proud of
We were happy to find a library that allowed us to accomplish voice recording. Once we accomplished this feat, our objectives continued to be met as we found a great speech recognition model and developed a nice UI.
What we learned
Embrace the challenges and continue to move forward no matter how much time is left. Although our main priority was to have a nice team project built, we realized the importance of deadlines and tailoring our objectives to meet those deadlines.
What's next for No-Attention
Easy way to add more books and transfer learning. Voice and accent learning.