Nuance, IBM partner for advanced speech recognition research

The world’s foremost expert in speech recognition technologies has teamed up with one of the world’s most respected technology research organizations in hopes of taking voice interface technologies to the next level. Nuance Communications (NASDAQ:NUAN) and IBM (NYSE:IBM) today announced a five-year partnership to collaborate on producing more natural and intuitive voice recognition technologies as well as a new generation of applications that utilize them.

Nuance chief scientist Vlad Sejnoha said the partnership was a natural fit: IBM Research pioneered synthetic speech and voice recognition back in the 1970s, while Nuance has taken the lead in speech technology over the last decade. Combined the two will have more than 100 researchers devoted to the project, spread among Nuance and IBM’s facilities globally, Sejonha said.
The project will tackle two major goals. The first will be to refine and improve today’s voice recognition, interpretation and text-to-speech technologies.

Leveraging the mathematical prowess of both organizations, they hope to create new, more sophisticated algorithms that can detect intonation and nuance in human speech as well as the basic meaning of the words spoken, Sejnoha said. For synthesized voice, they’ll develop more natural speech, which captures the flow and tone of the human voice rather than today’s more abrupt and automated synthesized speech. And they hope to create recognition engines that can deduce intention among multiple possible interpretations in an attempt to move voice interaction beyond basic commands, Sejnoha said.

By investing both their research efforts into the projects, Nuance and IBM hope to think far beyond the next round of voice-powered applications and focus on how voice will be a critical interface in networks of the future, Sejnoha said. “It allows you to look at the near term, but it also allows you to make some long-term bets,” he said. “These are the bets that are going to pay off.”

IBM and Nuance plan to pay particularly close attention to the handset interface, a platform that stands to benefit enormously from voice recognition technologies as it becomes more versatile and data-centric. Today conducting a simple task like a Web search takes multiple steps with touch-screen and keypad interfaces — opening the browser, navigating to the search engine and typing in the search terms — but with voice recognition and interpretation, they could be reduced to a single voice command. Today visual user interfaces are very stratified, Sejnoha said. “Speech can reduce the distance between you and what you’re looking for,” he added. “It allows you to cut through the tyranny of the user interface. You can talk to things that aren’t visible on the screen, for instance.”