Silly questions about telephony

Hello Internet, I have some questions.

I would search for them online, but, I haven't yet figured out the keywords I should be using. There seems to be a large overloading of meaning for telephony terminology.

Question one. From a program running on GNU/Linux PC, I want to have the capability to connect to a USA cell phone network as a voice call and send and receive audio over a cell phone voice call. The program will be generating the audio. Imagine a robo-caller program that would be calling a list of phone numbers to play emergency announcements.

Question two. From a GNU/Linux PC, I want the capability to connect to a USA cell phone network and make a voice call to an old-school 9600 bps landline audio modem, and have serial comms with this landline audio modem.

In both cases, the PC doesn't have any native capability to connect to the cell phone network, so it either needs a service or some hardware that will provide this capability.

I understand what I am trying to do may be silly. That is the point.

archive.org is good for old tech docs

I saw on Undeadly a note that OpenBSD's Ted was patching the ancient bcd program, which converts text into ASCII-art representations of punch cards. Punch cards were a technology from the 1960s and 1970s (?) that stored code or data on cardstock, with holes punched out of them. Each card held a line of text. If I recall correctly, each character was a column on the card, with as many as seven holes punched out of set of 12 possible locations. There were 40 to 80 columns on the card, according to the brand and the decade.

Anyway, Ted modified the bcd program to read in the ASCII-art representation of punch cards that it generated, so that it became, essentially, a very inefficient reversible encoding, but, he was unsure where to search for documents that he could use to validate the output.

My goto place for tech docs from the 1970s is archive.org. If you've never searched its collection, you should check it out.

Anyway, I did manage to find a couple of references there to punch card encoding.

For a couple of brands, anyway, punch card encoding seemed to have, for each character, a 4-bit "zone" or category and an 8-bit index. But this didn't result in a 12-bit encoding. Only a sparse subset of the available 12-bits indicated a character, for mechanical reasons, I guess. A subset of the characters now included in ASCII were encodable, but, missing some punctuation such as square brackets. It is for reasons like this that the C standard has trigraphs.