Proper implementation should be filtering its own audio output stream out of the assistant mic input as it gets reflected back in.

Apple definitely does this. The homepod is pretty great at picking up a voice with loud music playing. But then it goes to Siri...

I'm not sure how you would do that. The audio it hears will be very different than the original waveform it is playing.

Siri will listen when music is playing; this is quite handy actually because you can say things like "make it louder" or "stop playing music".

Or what song is this

Does that work on any iPhone? I’ve never considered doing it on purpose.

It needs to listen for the stop command.

I have an alarm that plays a song. I don’t want to listen to the whole song every morning so I stop it once I’m up and going. This is through Alexa though and Alexa repeats the song until you stop it.

It hadn’t occurred to me people might use voice commands for alarms. I press the button on my phone or watch.

I charge my watch at night so interact with my alexa 99.9% through voice.

I used to rely on getting out of bed to switch it off manually but over time found I didn’t need to do it.