I don't see any way a tone is better than a voice yelling at me in a language I don't understand.
Perhaps the brain has an easier time distinguishing between a reasonable number of distinct simple tones rather than between complex words you do not understand and can more easily be confused. If you have to deal with 2-3 alarms then tones are pretty intuitive, reliable, and very simple to implement even with century old tech. If you have 50 critical alarms then you probably want them explicitly informative, words in your own language.
Today the need for complexity and the fact that we have advanced, reliable technology for alerts already made voice warnings standard especially in complex systems operated by well trained individuals (pilots on a plane).
A loudspeaker is ancient technology and trivial.
And you’d need to ensure it‘a compatible with the letter of every law and regulation whether you are marketing your vehicle.