Future Work

Some potential avenues for our work include improving customer service interactions between customers and employee representatives and potentially providing those with hearing loss issues accessibility to more information when conversing with people. Extending our project to meet these goals involves a variety of different future improvements including training our model with more data containing a wider range of voices and accents. With more data, the model can learn different voice patterns and more accurately predict an emotion. We will also attempt to model more emotions than the 4 we focused on in this project in order to widen the range of applications for our web application.

In terms of our data modeling, we could potentially harness transfer learning to achieve a better classifier. There are pretrained models that have been trained with thousands of images such as ImageNet and therefore are robust in feature extraction.