Whoever came up with that wasn't entirely right in the head: so now people will stare at the device for another five seconds wondering what they did wrong. Sound of impact.
It's not impossible, just a very annoying way to control something precisely.
It's literally the worst way to interact with anything. It's infuriating how bad it is, especially for someone whose first language isn't English. It understands maybe 20% of the commands I give it, and the delay between clicking the button, it doing the "bong" to indicate it's ready, then listening to my command, then it processing and finally giving me a response(usually "sorry I didn't understand that") is just horrendous.
Not to mention that it's still physically impossible to use voice controls to dial people with non-English names if you have your OS set to English - try saying "Dial Jan Szczepanek" and it will never recognize it, no matter how you pronounce it. If you have OS set to Polish it will work fine, but I cannot be expected to keep switching my OS language!
This is just the modern allergy to allowing user configuration. (God forbid someone use your software to do what they want to do!) The problem was solved in the 90s, when speech-to-text sucked, by letting you define your own words.
https://www.speechrecsolutions.com/tutorials/Customizing_Dra...
Obviously, that solution will still work today.
But you are still allowed to configure your contact information, so there's an obvious solution: instead of calling him Jan Szczepanek, just call him John Stevenson. Everyone wins! ;D
However, it also has a "hey Google" like feature, where it's always listening and triggers voice control when it hears the words, without the push of a button. And it triggers randomly when I'm talking with my wife, even when we say nothing remotely close to the keyword.
Funnily enough, we also didn't know this feature existed, and neither did the dealership. So for a while we thought it was a random bug. Eventually, my wife figured it's a feature, and it can thankfully be turned off.
But for other stuff (changing tracks or albums - "play xxx from spotify") it works just fine for me. And yes, I use a louder "computer" or "command" voice when talking to digital assistants (had to learn how to do that to control Amazon's Alexa devices). ;)
Does that work with foreign titles? Dragostea Din Tei? 爱的就是你?
It's just not reliable enough for everyday use.