This sort of thing used to be non-trivial. I hacked this together probably in like an hour.

Ho boy, the times: they are a changin.

This is probably going to sound archaic in a few months, but a lot of "home assistant" type devices right now use a technique called "slot filling" under the hood. An "intent" classifier is the component that figures out what the relevant slots are and 'fills' them with values, resulting in a command being emitted and arguments passed. Instead of training one bespoke: you can probably just use this directly with no or very little modification.

1. git clone <this repo>; cd <this repo>
2. pip install -r requirements.txt
3. Create a file named .env containing one line: OPENAI_API_KEY=..., replacing ... with your key.

$ python main.py  "becca, how would I drive from my home to SeaTac airport?"
## {'intent': 'get_directions', 'arguments': {'start_location': 'home', 'end_location': 'SeaTac airport'}}

Act as the intent classification component of a home assistant, similar to Amazon Alexa (except your name is 'Becca', not 'Alexa').
Common intents include: play_internet_radio, play_song_by_artist, get_weather, current_time, set_timer, remind_me
You receive input in json format: {"input":...}
You respond in json format: {"intent":..., "arguments":{ ... }, }}
{"input":{spoken_request}}