Back in the per-Divestiture days, say mid 1970s, the Bell System has copper wires and black telephones. The wire supported up to 4 KHz of bandwidth and the repeaters were of the highest quality as were the speakers and microphones on the boat anchor phones. Voice quality was a sine qua non. You could whisper on the phone and the other person could hear unabated.
Along came several new "inventions". The first was speech compression. Namely we were using digital for signal;ling especially on the new cellular networks and we wanted to consume as little capacity as possible. Thus we lost the 4KHz analog to 1.6 Kbps compressed voice, I called it Donald Duck in a can.
Second along came IP, Internet Protocol, namely we used that for networking and ultimately for direct communications to the end user. IP is great for data but not so great for voice. You see packets traverse at different times and speed. Donald Duck starts to stumble.
Third, the small cell/mobile hone ad horrible microphones and speakers. In addition people would walk and hold them at a distance. In addition Customer Service people had these otherworldly headsets and microphones. mumbling across a compressed IP network.
Now we have to have a conversation where every other word is "what did you say?" or worse, the other person just rambles on incoherently.
So try this in a service where you add a language barrier. You see those codex compression devices were made for standard American English, no Brooklyn English and not Dominican Spanish.
Thus finally we have speech recognition, a true misnomer. Not only is the speech not recognized but even if recognized the system is not prepared to deal with the answer the caller provides.
I guess the old copper stuff was really good!
Oh yes, then finally add the mask thing!