I read that a wirewound (direct drive) speaker is 30% efficient, and a piezoelectric speaker is 60% efficient. I think it might be possible to make wirewound speakers 60% efficient. This extends battery life.
Ok, so think of a coil of wire around a magnetic broomstick; when you energize the coil,
the stick moves inside the coil of wire. If you fancy it up, and add an umbrella to the stick tip that moves a bunch of air and is a lot like a direct drive speaker.
But think of the magnetic coil! It still has its outer perimeter not attached to anything to drive. If you slipped an external torus around the outer perimeter you would get an audio speaker with a torus around the base. The torus would wiggle along with the music, but it is extra mass being driven! It uses more electricity to do that, but maybe not a 1:1 ratio.
Now what if you go all pottery wheel with the torus so it makes a big outer cone or frisbee around the core speaker; then you have two acoustic emitters from one electrical winding. So you you get more moved air volume from each full-travel of the speaker driver magnet. So it can sound louder.
Audiophiles might find this well intentioned but unappreciative of sound quality.
The way the cone/frisbee sounds compared to the regular speaker is adjustable at design time. That's where the genetic algorithms have value. If you define your fitness criteria as anything with better frequency response, then the genetic algorithm reshapes the cone/plate until it finds something that works.
A bunch of engineering things about the cone/frisbee like stiffness and shape, acoustic reflection, and beamspread can be geentic algorithm variables. The genetic algorithm sifts through a huge plurality of models of different shapes and stiffnesses and energy efficiencies. Size of frisbee/cone and regular speaker can change in relation to each other.
It seems possible this system could increase available loudness, and increase energy efficiency. Perhaps direct drive speakers could go up from 30% to 50%. efficient.
In fact, if you really want to save energy just make a mini 2x2 grid of smaller speakers; make them all fisbee+speaker duos, then just run the number of them (1-4) you need to use to get the preferred volume (and optimize frequency response just like running separate tweeters from midrange and bass).
Why would anyone bother? Airpods have a 20 hour battery life, doubling that to 40 hours might be possible. (30->60%). Also people use airpods instead of hearing aids so this could be a slightly medical beneficial technology as well.
I think $1 dollar store headphones, less than 7c each on alibaba are assembled by robots, so a duo frisbee/speaker is going to be similar in $. Even if the 2x2 matrix is used
(to save energy while while improving frequency response) that's just 21 cents more per earbud is used. So an alibaba version might be 42 cents. Airpods go for over $200 (2020).