Just some hacks to start a discussion maybe some tools to provide KWS word datasets from ASR sentences?
Just a start and no have to figure out some other processes to trim and center better.
Does Kaldi ASR do something similar and more importantly is it accurate with its inference timings?
https://drive.google.com/open?id=1wAb8SBwC02izZiGxO9dLVkRobiH_sYUg