Is thare any option provided to feed custom data to train the keyword detection model in azure speech services custom keyword creation. In the instructions document there is a screenshot with this option. But cannot find this while creating a model. And also is there any way to reduce the time to train the model. It takes more than 3 days. Can i solve the problem by upgrading the package?