What's the evidence showing it was gradual? Big jumps may be unusual statistically [what set are you sampling? 'planets with apex predators that are bipedal mammalians' looks like too small of a sample to make statistical inferences??] but then so is genesis and it seems that happened.
This expression is more symbolic than to be taken at face value.
> meaningful enough - OP
this means there's an optimum
> Given more animals that produce sounds, the ones that can transmit more information - OP
This means animals below (a?) optimum threshold. Shannon called that threshold the bandwidth, I believe.
> [complex sounds] take more energy to recognize
So this is the other side of the equation. We have a signal to noise ratio that limits the bandwidth. Energy is Information and Energy conversion, ie. Information transmission is inherently lossy, because Entropy decreases in closed systems. The trick then is to use the information to expand the frame of reference, to grow as an individual and as a species. That gives a lower bound on what information has to be converted into the system - that's what OP is talking about.
> What specifically will kill you if you transmit less information
In the limit, not transmitting any information means heat death. So, OP was talking hyperbole, obviously. The question for a local optimum is obvious - birds proliferate and not in a small niche. A global optimum in the infinite, infinite bandwidth at some point, or infinitely increasing bandwidth in infinite time are obviously out of scope - even if we like to dream on cosmological scales. Hence, OP's leading statement is rather symbolic than to be taken at face value.
Given an animal that can produce sounds, if sounds are meaningful enough it can be a big advantage. Given more animals that produce sounds, the ones that can transmit more information will survive.
It's weird to think that there's nothing between complex (human) and simple (sheep) when it was the result of a gradual evolution. Big jumps are rare.