As a side project I wrote a python program that uses Google's Speech Recognition API to loop through segments of any audio file to extract words to a text file including functionality to restart the process at the correct location in case of network failure. The idea is later to apply some sort of sentiment or recommender analysis to determine subject matter and tone of conversation. It's a good idea, but unfortunately the speech recognition API runs at natural speed. There may be a way to accelerate it, but that will have to be something to consider adding in a future release.
I’ve also been in conversations with an individual in Denver on a startup idea involving knowledge graphs and natural language processing. Conversations are ongoing. That said, combining CUB classes with my MOOCs in Python and C++ — and now my ongoing revisitation of Andrew Ng’s Coursera based Machine Learning course — I am hard pressed to meet my goal of at least one blog post per week! I’ll have a lot more to talk about soon!