How to train a different model for speech command example of tensorflow lite for android

Question

I was trying to train my own model for speech command example provide by tensorflow for android(https://github.com/tensorflow/examples/tree/master/lite/examples/speech_comma