Tbf there isn't much in the voice activated smart speaker space. If I could just program some speech to text abstraction and link it to automations I would. None of the solutions really seem mature enough.