bug13 said: ↑
...I image this info can help me better understand how to minimize the feedback..

Well spotted wayne.

I also note that this is for an intercom, which has both a speaker and a mike the recipe for feedback.

The usdual way for intercoms to avoid feedback is the function switch, either listen od speak, but not both.

If you must have both then both the mike and speaker need to be highly directional.

The use of a circuit like this can also be pretty effective at reducing feedback problems with 'full duplex' intercoms:

http://www.ecircuitslab.com/2011/06/full-duplex-intercom-circuit.html

Not as effective as 'press to talk- release to listen' switch of course, but can work quite well as a hands free intercom. Often used in door intercom circuits.

Note that you need 3 wires between the units. So it's no good as a 'speakerphone' circuit, which needs to work on a single pair....

 

They make automatic feedback suppressors for sound systems such as these, which have multiple programmable digital filters that automatically add a steep notch filter to the feedback frequency(s).

 

As a final piece of confusion, bear in mind that the average human voice, when speaking, is between 85 and 255 Hz, harmonics notwithstanding.

Don't let these fundamental frequencies limit your design. I know faley did qualify his statement by including the phrase "notwithstanding harmonics". Not to add to any confusion, but telephones are designed for 300Hz to 3.4KHz, with resulting reduction in intelligibility. Think of how it is difficult to distinguish between t, b, and d on the phone.

Human hearing can distinguish from 20Hz to about 20KHz, with most sensitivity between 300Hz and 10KHz. Hence the phone upper limit of only 3.4KHz is the reason for the muddy t's.

I ran into this when designing my audio animatronic controller. My reference for this post was here.