ethan - 2018-12-30

I am reading the latest (2016) PNCC paper. https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7439789

There are too many blocks to understand in it and I am getting lost in the explanation.

Specifically I want to know How and Why the medium time power calculation is used to motify the short-time power calculation.

It says in the paper (page 7, left column, bottom) that

The time-averaged, frequency-averaged transfer functionS˜[m, l] is used to modulate the original short-time power P[m, l]

I don't understand why do they use the term modulation. I thought they were going to subtract the noise power from speech signal power, why are the multiplying the two?