Would anyone like to help create an open wake word model?

[sangheonEN]

Is there anyone who can help me create a voice wake word trigger model by referencing https://github.com/dscripka/openWakeWord GitHub?

my fork github repo is https://github.com/sangheonEN/openWakeWord

I'm at a certain level, but the model training is not working properly, so the prediction value is at the level of 0.0001, and the model doesn't work at all. I'm trying to analyze the cause and improve it, but I'm alone, so I can't figure out a direction, so if there are people who have the same concerns, it would be good to talk about it.