I’m not sure whether the author was being sarcastic with the headline “When Edison chose not to invent speech-to-text tech”, but in case they weren’t, invention is not 1% inspiration and 99% just deciding to do it…
By 1911, Tesla was indentured to J.P. Morgan any ideas Tesla had belonged to Morgan. Edison was financed by Morgan as well although enjoyed a greater amount of latitude then his counterpart Tesla.
I am having more fun imagining how Alan Turing would have gone about it. His first problem would have been storage and cache. He could play voice samples from phonographs, possibly cylinders with continuous parallel tracks, so he has a signal to compare the incoming signal with but then he need to do some sort of real time fast Fourier transform on both inputs, and compare the outputs.
But then maybe do the FFS once on the samples and bake them in to the comparator with patch leads. He still has a technical problem of storing up to a second of audio while the transform is done.
I am fairly confident that Turing could have got it working, but no, probably not Edison or Tesla.