It has been at least ten years since I worked on this stuff, so if any of it matters to you, please do not rely on my memories.
When I was working for a voice messaging vendor, we were part of an industry initiative to develop a protocol for passing messages between messaging systems from different vendors. The relevant specification was for the AMIS Analog protocol (and I've forgotten what the acronym AMIS stands for).
Basically, the protocol allowed a user of one voice messaging system to address a message to a user of another voice messaging system (the user interface was left unspecified -- that was a matter for the individual vendors to handle), in such a way that the voice messaging system could then dial the recipient system, do some handshaking, and then deliver the message in a way that allowed the recipient system to deliver the message to the intended voice mailbox.
The protocol relied on use of two fourth-column tones (C and D, as I recall) to screen out nearly all accidental calls to the incoming AMIS Analog phone number (I forget if the specification required it, but we came up with a canned message to play if that phone number received an incoming call that did not send the correct tone -- just to be polite).
It was quite a change from our usual workday (writing documents or code, and testing code) when we got to test with other vendors. We actually got to talk to engineers working for our competitors, and send each other messages (mostly we sent protocol errors, actually, to verify that the error handling on both sides was working properly - it does not take very many correct messages to verify that things work properly). I still remember one of our competitors (who really should have known better) who had the wrong country code for the US - but otherwise, there were not many problems getting things to work.
Security was an issue -- once the testing period was over, our management refused to allow an incoming telephone number - so other members of the AMIS committee could not send us AMIS Analog messages (rather frustrating, but out of my control).
I had a four column analog phone on my desk for a few years while I worked on this system (the fourth column, if I remember correctly, was to the right of the normal three columns, with the rows 1-2-3-A,
4-5-6-B, 7-8-9-C, and *-0-#-D). When I was doing it often, I was able to manually imitate a voice messaging system, computing checksums as needed in my head (there were not very many of them, and only two were variable, if memory serves). The protocol had to be designed to handle some slow systems, so I could dial fairly slowly when I was adding the checksums.Mark