There has been some writing on how it used wave guide effects to transmit, but not so much on the microphone part. Some of you may know that many speakers can be wired up as microphones. One wonders whether a device intended to be a microphone, might in some circumstances act as a speaker as well.
If the Cubans are following an evolutionary path from the device concealed in the Great Seal in the 50's then one could easily imagine something built into the structure of the building itself that is accidentally generating sound energy.
As a thought experiment, imagine that you could build a window frame that would pick up sound from inside a room, shift it to an inaudible frequency and then emit that sound. Furthermore, what if a double pane window intended to block sound from leaving the room, had a pickup attached to the inner glass pane and emitted the high frequency sound from the outer pane so that an external laser could pick it up. Then further imagine that the window was installed wrong way around so that the constant street noise was broadcasting high frequency sounds into the building.
I expect that the Russians don't do this stuff anymore because they get better results from HUMINT, especially deep cover agents.