So we’ve already concluded there’s something amiss in their power and battery life statements.
You’ve shown at your chosen volume level it’s at last 20 hours proving it’s possible so not invalid claim.
But you’ve also shown most probably the batteries are not as much as 13400 in terms of releasable energy.
So my current best guess is that the cells lower their Watt-hours as they discharge like all cells do, so if you play at a higher Wattage - more sound - then the releasable energy is less.
Say you could 20 hours at 20% volume and cells cut off having used 80% of their energy. They cut off at 9 hours at 40% volume having used 70% of their energy.
The harder you work a cell the less you get from it. Louder speaker is working cell harder.
It would take someone with a USB meter to know more but as an example based on assumptions.
You showed at your chosen level of volume 5V 2A 10W was 5h40m.
Assuming as they state it needs 5V 3A 15W that your 10W was a straight flat line (no trickle) then you injected 50.7Wh.
Assuming charging 90% efficiency the cells got 45.7Wh, of cells capable 96Wh so the cells discharged 47% of their energy.
Assuming the max volume of 80W speaker is say 30W (no distortion) and assuming 90% efficiency then the 45.7Wh would give 41Wh to speakers drawing 30W so battery life is 1h22m. That’s assuming they cut off power at the same Watts, actual would be lower if higher Watts made it cut off at a higher Watthours drained.