My Presentation at the Massachusetts Junior Academy of Science Symposium

I was thrilled to present my project, ‘Machine Learning based Speech Recognition App for individuals with speech disorders’ at the Massachusetts Junior Academy of Science Symposium. This project won the first prize in Massachusetts Science and Engineering Fair, $500 Foundation Medicine award, US Air Force Academy Award and US Naval Science award. It is very close to my heart, since it is a step in a direction that helps me and many other people like me that have a speech disorder.

I prepared for the virtual presentation and the questions. I was never sure what could be asked but I knew that I had to be ready to answer as fast as possible. So, I practiced and practiced.

Once the presentation finished, I was sure I did good, but I was still nervous. It is a tough competition, one has to be nervous, right?

Two days later, I got an email saying that I came second in the symposium. YAY!!! I was so excited. The comments from the judges were also so encouraging and motivated me to work on the project more so that it can impact the lives of many more people with speech disorders. I thoroughly thank them for the feedback which is below.

First of all, we would like to say that your presentation was quite clear in communicating the problem you wished to solve, your approach to solving said problem, your implementation of your solution, and next steps. In a problem with something like machine learning, this is difficult to communicate, so we applaud you!

Second of all, we believe that the problem you had chosen is extremely impactful. While you have implemented a first attempt at classifying your own voice, we foresee an application to many people who have speech disorders. 

Third of all, we believe your approach to solving the problem is a great first start! For some critical feedback, we believe that considerations about scalability for application of this idea to people with varying speech disorders would be necessary. With any type of machine learning model, it is necessary to have a large amount of data, and different people with speech disorders may require a completely new set of data. With this in mind, considerations of ways to mitigate this need for large amounts of data might be useful. A useful method for streamlining data collection would also be useful. Additionally, with the data that you replicated, adding a bit of noise to said data could be beneficial to making the new samples unique enough that it benefits the training of the model you have.

Understand that you are doing amazing work, and at the age you are now, we cannot imagine where you will be during your undergraduate or graduate degree, if you decide to pursue one. Congratulations, and keep being curious! It is paying off.