IBM Watson’s Text-to-Speech Gives Hotel Guests a Helping Hand
IBM made headlines a few years ago when IBM Watson was a “contestant” on the television game show Jeopardy!
Various Watson capabilities have continued to evolve, including Watson text-to-speech (TTS), which is finding its way into hospitality, educational, and call center environments, making IBM a 2016 Star Performer award winner.
Other types of enterprises are bound to eventually make use of IBM Watson TTS, though it is in its early stages, with more complete speech applications still on the horizon.
“We’ve broadened the Watson cognitive platform to increase its ability to understand, learn, and listen,” says Michael Picheny, IBM senior manager of Watson multimodal. “We can connect it with a large number of cloud-based APIs to help people find different ways to achieve their goals. The way people are using speech is really expanding, from conducting simple banking transactions to finding information on healthcare plans to other uses.”
One of those expanded uses, as noted, is hospitality. IBM and Hilton Hotels and Resorts are piloting Connie (named for hotel chain founder Conrad Hilton), a digital concierge that can manage some of the simpler guest requests, enabling human concierges to handle more complex issues. Connie is designed to augment, not replace, human concierges, Picheny stresses.
“Connie will greet [guests] in the hotel lobby, and you can ask it questions, about the hotel, its hours, its operations...or just shoot the breeze,” Picheny explains.
Rather than monotone robotic speech, Connie leverages what Picheny calls emotional IQ to produce expressive speech, so it is “apologetic” in tone if it can’t provide the information or the service the guest requests, rather than the upbeat “I’m sorry,” that some IVR systems (for example, when account information can’t be located) use.
“This is important because it provides the feeling that Connie has some knowledge of what is going on,” Picheny says.
The more guests who interact with Connie, the more it learns, adapts, and improves its recommendations. The hotel will also have access to a log of the questions asked and Connie’s answers, which can enable improvements to guests’ experiences before, during, and after their stays.
Connie uses a combination of Watson application programming interfaces (APIs)—including speech-to-text, text-to-speech, and natural language classifier—to enable hotel guests to ask about hotel amenities, services, and hours of operations. Through an interface with WayBlazer, a cognitive travel recommendation engine that uses IBM Watson and proprietary cognitive computing technology, Connie can also suggest local attractions outside the hotel.
There are some areas in which Connie cannot handle some of the same functions as a human concierge, Picheny concedes. For example, it cannot order tickets for different events because it won’t accept payment card numbers due to privacy concerns (letting others in the lobby hear the guest’s credit card number).
IBM is in the earliest phases of adding human-like movements to Connie’s arms, enabling it to more closely mimic human gestures, according to Picheny.
Also still in its early stages is a joint effort by IBM and the University of Michigan to develop a new class of technologies designed to enable people to interact more naturally and effectively with computers. The $4.5 million venture is code-named Project Sapphire.
The first step is a digital adviser that is designed to augment, not replace, human advisers for university undergraduate computer science and engineering majors. At this stage, the digital adviser is helping students register for classes, find degree requirements, and similar tasks, according to Picheny. Currently, the interface is text-based, with speech-based capabilities expected in the near future as IBM and the university continue to develop their speech-based technologies and interfaces.
IBM Watson is also being used by iQventures, a contact center solutions company, to provide business clients with call center analysis information, a program that started earlier this year.
iQventures’ Speech IQ transcribes incoming calls through its own automated speech recognition engine and runs these transcripts through Watson’s language sentiment tools, which in turn use text analysis via natural language processing to help to understand sentiment, keywords, and high-level concepts of conversations, helping to maximize call center effectiveness.