Those of you who’ve been listening to the podcasts I’ve been doing and reading the blog will notice quite how excited Cisco’s Jeff Rideout is with the Health Care Interpreter Network, which is a video over IP based network currently being used to share translation services in several safety net hospitals in California. Of course moving human translators by the magic of video telephony is a great advance over waiting for them to show up from wherever they are, and when providers have to deal with patients speaking as many languages as a typical Bay Area, LA, New York or similar facility sees, avoiding getting lost in translation is pretty critical.
But of course this being America there’s another way, and it too is pretty damn clever. I was shown a demo by a company called Spoken Translation which has an English-Spanish automatic medical translation tool. Here’s a cut from their blurb (as I’m too lazy to describe it all) but I’ve seen it and it works as advertised.
Converser for Healthcare provides 24/7 live interpreting. Its initial product will be targeted to the healthcare market and is a Spanish to English, English to Spanish translation. The system allows people who do not speak the same language to hold broad health-related conversations in real time, without a human interpreter.
Converser represents a fundamental advance in Machine Translation (MT) technology. No other system on the market today can provide reliable, bi-directional, real-time, wide-ranging translation via multiple interface modalities including speech recognition. Never before has a commercial product for conversational translation enabled a user to verify in real time that the translation is accurate, and, if not, to correct it on the spot. While Spanish is first out the door, other languages are planned for release later this year. Chinese is planned for the healthcare market, while German and Japanese are currently under development for other markets. Converser can run on Tablet PCs or laptops (full-size or ultra-portable), and release is planned for numerous handheld devices.
The tool allows input via typing, onscreen keyboard, voice (using Dragon Nat Speak 9) and handwriting recognition. It’s a touch cumbersome in that of course each phrase needs approval and potential correction (in case of possible English/Spanish ambiguities) but it appears to be damn clever, and at $1500 a seat, a darn site cheaper than having a human sitting around.
Now, if they just make one in Arabic, that’ll solve a wee problem the US military has in another part of the world.