Some major EQ would likely need to be applied. Assuming you're trying to bounce the laser directly into the detector, the biggest challenge would be alignment of the laser and the photocell, which would restrict your choice of positions, especially if you're trying to remain undetected. Modulation (the excersion of the beam at the detector) and beam width would of course increase with distance, but would also depend on the point of aim on the window, the angle and the size of the pane.

There was talk of this technique many years ago. Some TV report was done on it, but they got far less audio quality.

How about you bounce the laser onto the ground or onto an adjacent building wall, etc., and use your telescope to detect the modulation from just about any one position at a distance? Or is that what you had in mind? A photocell tuned to the the laser's wavelength would help the sensitivity. Maybe you could do that with simple filters, as on the TV remote sensor, being as they have already worked that bit out-- just need a laser to match the passband. Or do you even need a laser? Maybe you could use an IR spotlight. I'll expect a working prototype this weekend. Carry on. In the mean time, keep your super-secret decoder ring set to "Uniform".