The goal of this project is to train a neural network to distinguish and classify the accents of English speakers from different geographic and linguistic backgrounds.
The motivation behind developing a model to recognize accents in spoken English is primarily two fold.
First, if it is possible to determine a speaker's geographic location or native language simply by their accent, then it might be possible, for instance in a call center, to more efficiently route that person to a regional representative or to a speaker of an appropriate language.
Secondly, accent recognition is simply a necessary precursor to automatic speech recognition (ASR), such as is found in Siri-to understand what a person is saying, there must be a model in place that expects how they are going to say it.