Jibbigo Speech Translator Technology Developed at Silicon Valley Campus-Silicon Valley Campus - Carnegie Mellon University

Jibbigo Speech Translator Technology Developed at Silicon Valley Campus

Alex Waibel
Alex Waibel

Jibbigo LLC, a start-up company founded by Alex Waibel, professor of computer science and language technologies at Carnegie Mellon University, has released an iPhone application that translates English and Spanish speech back and forth.

The Jibbigo app, which has been out in the market for five weeks, has already landed a spot in the top three applications in the travel applications section of the iPhone store. The application has a word bank of over 40,000 words, in the domains of medical field and travel. This first release was designed with tourists and medical workers in mind. “The implications of this new technology are tremendous. Currently, Jibbigo’s domains include travel and medical assistance. Imagine how we can bring the entire global community closer together when we address new domains, like humanitarian aid, education, and business,” said Waibel.

Dr. Alexander Waibel is a Professor of Computer Science at Carnegie Mellon University, Pittsburgh and at the University of Karlsruhe, Germany. He directs interACT, the International Center for Advanced Communication Technologies at both Universities with research emphasis in speech recognition, language processing, speech translation, multimodal and perceptual user interfaces, which now has a new center at the Silicon Valley campus.

Waibel’s team of researchers at the Silicon Valley campus worked on the statistical machine translation technology that is used in the iPhone app. The statistical machine translation approach has the advantage of rapid deployment and higher accuracy than traditional methods, however, it is very demanding of computing resources. The Jibbigo team at the Silicon Valley campus was the first to successfully deploy a full-scale statistical machine translation application on a mobile device.

“In the past, people used grammar-based translation techniques, which requires human efforts in writing grammar rules to build up a translation lexicon. However, this is expensive to build and difficult to deploy for any new set of languages and domains. By using data-driven machine translation, we are able to learn the translation model automatically from a set of translation examples. This allows us to create a new set of systems for new languages or new domains in a much shorter time frame,” said Joy Zhang, assistant rsearch professor at Silicon Valley and machine translation expert.

“We wanted to bring Jibbigo to Silicon Valley because we see the potential for numerous partnerships with the myriad of technology companies here. With the Silicon Valley campus being the new home of Jibbigo, we will be able to facilitate these interacts much more fluidly,” said Waibel.

For more information, visit http://jibbigo.com/.