Speech recognition continues to get better and labor cost keep rising, but, as an engineer in habit of jumping 'out of the box' like Jack does, I like to think about alternative solutions. Here is one that is amusing.
I walk into a little corner restaurant in Paris to have lunch. As the waitress comes over, I flip-open my cellphone and press a button that was programmed to connect to a translator service which was part of my vacation package. When the waitress opens her mouth, I point the phone to her.
Francine is sitting in front of her computer and writing into her blog. She lives in Chicago and works part-time as an on-demand translator. When her computer beeps and pops open a window, she is looking at the waitress opening her mouth to say something. She takes a quick glance over to the side of the popup and sees basic info on the client. His name is Don and his location is in Paris. Based on his GPS location, he is in a restaurant. Francine proceeds to help Don order a lunch.
I picked this scenario because, the last time I was in that siutation, I ended up ordering a lunch of just side-dishes. It involves more than speech recognition, but the core idea is that speech recognition does not manadate machine doing the recognition.