Cognitive Rooms: Connecting People and Technology with Speech
The idea is simple: a person walks into a hotel room and says “Turn the temperature to 26 degrees” or “Book a table for two at the hotel restaurant” and the room’s voice responds, either by complying or asking for additional information. This latest developments in cognitive rooms just got a whole lot smarter.
In April this year, Harman audio and IBM Watson unveiled their Voice-Enabled Cognitive Rooms concept. The system works through concealed microphones and speakers linked through the Watson cloud and IoT services.
The idea is simple: a person walks into a hotel room and says “Turn the temperature to 26 degrees” or “Book a table for two at the hotel restaurant” and the room’s voice responds, either by complying or asking for additional information.
The concept of rooms that respond to our requests can be applied to medical facilities, corporate offices, hotels, cruise ships and other hospitality environments. It helps first-time hotel guests or hospital visitors achieve what they want, without having to find out where the information is stored in each case. The person is connected to the environment through voice alone in what’s known as a “highly connected experience.”
These voice-enabled cognitive rooms also function as an in-room concierge that can answer general questions or site-specific questions developed by the facility and featuring custom answers created by staff. Checkout times, gym hours, and hospital visiting times can all be supplied.
Systems with the ability to ‘learn’ over time
The solution isn’t one-size-fits-all, and it develops over time. Because Watson is a cognitive system, it can be trained with a body of knowledge that is ever-expanding – constantly adapting to changes in the environment and providing new solutions with the most up-to-date information.
And Harman isn’t the only audio company that sees the potential in matching Watson up with an audio feedback system. In January, Panasonic and IBM announced they were working on a concierge service to assist people when travelling. The idea is to create a seamless experience, whether you enter a hotel in Bangkok or Brisbane. It also allows hotels to upgrade traveller experiences in the competitive hospitality market.
Like the Voice-Enabled Cognitive Rooms concept, Panasonic’s Digital Concierge applies Watson and IBM Cloud to a Watson-enabled mirror installed in hotels and other hospitality industry customers. Using the digital mirror, hotel guests can have a spoken dialogue with the Digital Concierge to obtain information about the hotel and its services, entertainment and shopping opportunities, transportation, weather and other topics related to their stay.
“Panasonic has identified a need for this and several other kinds of connected solutions in the hospitality industry,” said Yasuji Enokido, president of Panasonic Corporation’s AVC Networks Company.
“Working with IBM, we plan to further implement our connected solutions vision while making use of Watson intelligence to provide end-users with more natural cognitive functionality as well as richer feature sets.”
Languages, accents no barrier for Watson
The key to the digital concierge/cognitive rooms concept is Watson’s ability to process unstructured data, as opposed to structured data.
Structured data is the type computers traditionally process. For example, a person’s weight, height, hair colour and occupation can be put into specific formats: 87kg, 172cm, blonde and electrician. Producing a list of all blonde electricians weighing 87kg, for example, is the kind of structured, clearly defined request that Google’s search engine processes millions of times a second.
However, Watson processes unstructured data, the type we exchange in everyday conversation, such as, “What time does the restaurant open?”, “I’d like to order some wine” or “Where’s the nearest supermarket?” Each of these elicits information pre-programmed into Watson. The tricky part is that Watson must recognise all the permutations of requests that should trigger that particular response.
The amazing thing is that to function properly Watson needs to filter out the unnecessary parts of communication to reach the underlying request. Voice pitch, accent and background noise must be removed to lay bare the data structure beneath, in the same way that an analogue waveform must be sampled to record its essential data as a digital file.
And considering that the rules of one language can be learned just as easily as the next, a voice-enabled cognitive room can understand and act on the request, “Send up an engineer to fix the shower” whether it’s spoken in Dutch or Hindi.
It’s a massive task. According to IBM, 80% of the world’s data is unstructured and 80% of that has appeared in the last two years. And data is not meaningful in isolation – it must be interpreted in a context, which is what Watson does so well.
If cognitive rooms become the norm, every service-related building will be expected to have one. Microphones tuned to listen for the human voice and high-end speakers that can produce a rich, realistic vocal response will be in demand. The implications for audio equipment designers and sales are tremendous.