>>506102
There are some problems I see, here.
First is that most microwave spectrum emissions are readily absorbed by water. Stick a ham in a microwave and see how the outside of the ham heats up relative to the inside.
Now - it is true that FM modulations could potentially be detected by the target, but this is more likely as pulse-width modulation. You use a base frequency in the microwave band and modulate the duty cycle, as opposed to the frequency.
This leads to an induced amplitude modulation at the audio frequency. Whether this could be felt by people is debatable. The tech for PWM at microwave frequencies is within the civilian market. I have other projects I am working on at the moment, but it is worth considering that microwave exposure is considered rather serious in the military. Particularly to the testes and cornea of the eye. Even modest RF leaks in test equipment are blamed/suspected of impacting reproduction (those of us in the RF end of things were told to expect to never have boys without going on leave for a couple weeks).
How much stock to put into those claims... Not sure. However, I would not recommend blasting people with homemade microwave equipment to see if you can make them hear voices.
On the other hand, using piezoelectric panels in the ultrasound range can use planar array beam steering algos and pwm to direct sound to a single area. Using constructive and destructive interference, it is even possible to position the audible 'box' within 3d space. Though how reliable this would be with echoes and other things in the field is questionable.