The energy of a shock wave in a structure will bounce around until it's absorbed. Structures for the most part are all about carrying vast forces gracefully, so unless someone threw a shock absorber into the design (like they do in Japan for earthquakes), then the shock wave dissipates in an unintended manner. Carpets will absorb energy, breaking glass will absorb energy, water pipes breaking, people getting knocked off their feet, desks and potted plants and 50-ton presses, everything absorbs some energy of the shock wave. But what doesn't get absorbed bounces back and forth through the structure. Mud would absorb lots, but the shock gets reflected at the basement due to impedance mismatch, and again at the bathtub itself, same reason. So breaking concrete at the top remaining intact floor (the one that's already under the most stress) is probably the biggest shock waves energy sink of the list.

The reason that sound (shock wave energy coupled to the air) isn't a big one on that list is the impedance mismatch between the steel structure and air, but all the flat surfaces of the structure act like the diaphragms of speakers, which tend to serve, unintentionally, as an intermediate impedance matching element. If not for glass and flat walls, you might not have heard any BANGs.

As for "metal hitting metal doesn't sound like a BANG", please note how I described it before, and again, here. A structural element that fails under load transfers an impulse to the structure. A truss is under tension so long as it's functioning. That tension is being constrained at both ends by an equal and opposite force. The opposite force at each end of the truss is stored in the deflection of the curtain-wall/core like a spring. (I wish I could come up with a better example, but I'm late for work and haven't had my coffee.) When that truss fails, both the curtain-wall and the core are no longer in equilibrium, each no longer has the truss' tension holding it in that deflected position (delta-s could be maybe a cm, maybe an inch), so each instantly recoils to its nominal position. This sounds like BANG to everyone there and everyone in the structure and outside because it's a big flat thing that's moving real fast in one direction, as if someone had hit it with a really big hammer or an explosion, which also would drive it one way, fast.

That shock-impulse travels down the structure in two modes, compression at 20,000fps, and transverse at a slower rate that's determined by the mass of the floors and the transverse strength of the structure, so below, you hear the BANG! first, then the rumbling of the transverse wave interacting with the various unintentional resonances of the structure. People at street level will hear the BANG from the structure (Mach 20) before hearing the direct sound of the truss releasing traveling at Mach 1.

Please note that while I used a failing truss as the example, virtually every structural element is either in tension or compression, and the human ear can't tell the phase of sound (it just doesn't work that way), so a tension failure will sound like a compression failure, and they both go BANG!. Let me know if you can think of a structural member that wasn't in tension or compression.

The myriad reports of random-spaced BANGs from witnesses make perfect sense in a failing structure, and make no sense in a CD. What's that "C" stand for? Why would there be random explosions at random places and times throughout the structure if the Demolition were Controlled?

It should be noted that what happened in reality was neither a pancake nor a bagel nor zipper nor any other pure-bred clean process, but a complicated mishmash of events, no offense intended against those who've made sophisticated models in good-faith. I think the strength of the non-CD argument is that all versions show that once the collapse began, the rest was inevitable, regardless of the details of the collapse mechanism.

Assume that the upper-section floor fragmented into small chunks and rained down gently to the floor below such that the kinetic energy didn't play a role in the collapse. The weight of the debris would still exceed the load-bearing capacity of the impact/fire-damaged floor, and it'd fail when the max-capacity load was exceeded. The floor below would then have even more debris falling on it, so it'd fail, too, and so on all the way down.

Let's say that the upper floor fell in larger slabs, but one at a time, so that the whole of the floor's kinetic energy was spread out over time. Each slab would try to pancake on the floor below, but there was office furniture in the way, some of which was more substantial than others. The more substantial stuff would've concentrated the load from above (the slab weight) onto the floor below through the smaller footprint of the desks/filing cabinets, etc., so that would load the intact floor unevenly, creating stresses in the floor that would've contributed to that floor breaking into slabs, itself. Falling slabs create falling slabs, and so on, and that progresses to collapse, too.

Big chunks likewise hit hard and make other big chunks.

Bageling is the next logical step in this sequence, but it isn't the only one, nor is it more likely than the Big Chunks Make Big Chunks model, which I believe is the likeliest, at least for much of the collapse regime.

The Small Chunks version has a hydrodynamic loading effect that's worth thinking about, too. That would tend to crush inward the core and splay-outward the curtain walls, creating a zipper-effect dropping floors below the one in the crush-zone. You actually have leverage working to pry the perimeter columns away from their trusses. This would tend to leave the perimeter columns intact and splayed outward, exactly what you see at ground-level.

I've got to get going.

Oh, aerosol cans - They're pressurized with hydrocarbons these days, and statistically, there are enough of them around that you'd expect one to be going off someplace in the structure at some frequency, and I dispute the timing of the flash being at the "start" of the collapse. The video showing the closeup of the curtain wall buckling in illustrates the point that the collapse didn't start instantaneously, at one precise moment, like you get from the detonation of a high explosive, it started like a tree starts falling over - slowly and with increasing speed, but so slowly that it's not a trivial matter to say when it started. Before you call that flash an explosion, please look at some of the videos of what real explosives look like when they go off. The eruption of smoke associated with that flash wasn't big enough to do anything significant to the structure, and real explosives don't produce long-duration flashes and big slow puffs of smoke, they go off in microseconds and produce supersonic shock waves. You get small pops like that in every big fire. It's something cooking-off. My guess is that it's an aerosol can. The site of that flash doesn't look that strategic to me, and the room was burning before the flash, so an aerosol can cooking-off seems the likeliest explanation.

Also, recognize that at the time that flash went off, the perimeter columns were already bent in across the whole of the face seen in the posted link, and when they bent in, they did so in unison. You just can't do that with explosives.