Real Time Live Mic Stream Inferencing?

[MXGray]

@Kyubyong Thanks for your cool implementation! I hope you can let me know the relevant input and output node names needed for freezing the model and my training results as a PB graph? I'm thinking of porting this to C++, or just using it for inferencing live mic streams through PyAudio or Speech_Recognition? I've tried experimenting with write_graph, and also with converting the PBTXT to PB, but no luck. Your help would be greatly appreciated. Thanks!