My experience in computer automation:

I was wiring large computer controlled machine tools (early CNC) for Burgmaster Corp in Gardena, CA in the early '60s. They were controlled by discrete-component Hughes and GE “computers” programmed by punch-tape with teletype I/O.

In the '70s and '80s I installed computerized building automation as a captive-contractor for Powers Regulator Co… This work typically consisted of interfacing all the motorized equipment (fans/air-handlers, chillers, pumps, etc.) in a hospital complex (City of Hope, Big Bear, Hemet, Marten Luther Anaheim,…), University (UCI) or municipal building (Riverside city hall and exhibit hall, City of Industry, …) to a discrete-component mini-computer through field-cabinets located throughout the buildings. The computer would sense the power usage in real-time, and as the maximum power preset was approached, it begin load-shedding by cycling the loads on and off for brief periods, thus reducing the peak-demand, used by the electric utility companies as a factor by which the total kilowatt-hour usage was multiplied to arrive at the price of the monthly utility bill. Twelve years of this work gave me a good background for computer control of external equipment.

My cover article published in the February 1983 issue of Computers and Electronics https://kj6yvt.com/speak&spell_Feb1983.PDF reveals how to access the words in the Texas Instruments Sepal & Spell with a simple computer interface. It even implements a logic-analyzer to expose the proprietary control signals used in the S & S. Although several articles dealing with the subject appeared in the iconic Byte Magazine, unlike mine, they all failed to actually enable a computer to speak.

Later in the '80s and '90s I was an independent computer consultant installing local area networks and providing Unix system administration services to small businesses.