What is the complexity of the implementation of the caller request recognition service and the pronunciation of the result by the robot?

Question

Speech recognition

I wonder what is the complexity of organizing such a thing at home, the essence of which lies in the fact that there is a separate telephone number, either landline (city) or mobile.
Anyone can call, to which a greeting audio file will be played in response, after which the voice of the caller will be expected with a request, for example Super-window store on Lenin , a certain program will recognize the request approximately and make a search in its database, and pronounce the results in a voice, ala
Super-window store
Address Lenin street 10
Working hours ...