Detection of Blacklist Speakers from Speech - MCE 2018

There are three reasons we are caring about this challenge:

• Long term Inorganic transition via medium term Organic Shift - We have always been strong at speech recognition and synthesis. Given the advances by industry, this might be a good time to diversify a bit into other domains of speech processing. This challenge presents a way for us to do this in an organic way: Accomplishing tasks in this challenge can be seen as first steps towards antispoofing and fairness in AI which we believe is an important problem to address in near future.


• Testing the 'algorithms' of our trinity - We subscribe to the notion that progress in deep learning is powered by three inter-dependent things: Data, algorithms and infrastructure. The nature of this challenge lets us test strength of our algorithmic pipelines.


• Incorporation of conversational agents - We plan to follow 'conversation agent first' approach to all of our AI based projects going forward. This marks the first project in that sequence. We will start with a chatbot for this. In other words, this web page will always be accompanied bby a virtual assistant that answers questions about the project. It can do small talk too. Say Hi :)