Inspiration
Both me and Syed had to attend speech therapy when we were younger. As we were brainstorming ideas, we realized speech therapy was currently extremely expensive, time consuming, and inefficient. We concluded machine learning could be applied to solve this prevalent problem.
What it does
Stutterless acts as a virtual speech therapist for people who stutter. Essentially, it provides realtime feedback to users and motivates them to reduce their stutter by showing them their progress.
How we built it
We used Swift to build the app and Azure to visualize data, eventually relying on CoreML and CreateML to create models.
Challenges we ran into
Creating a model took most of our time; getting a lot of data for people who stuttered was incredibly difficult. No data source had words of people stuttering. In addition, deciding on what type of data to feed the model (phrases versus words versus sentence) posed a serious problem. Finally, porting the model over to an iPhone app and allowing for realtime feedback also posed a challenge.
Accomplishments that we're proud of
The machine learning model and the user interface
What we learned
The type, variation, and source of data is incredibly important for machine learning.
What's next for Stutterless
We plan to collect more data to improve on the models and create a more comprehensive reward system for users using NLP anchoring.
Log in or sign up for Devpost to join the conversation.