On August 15, Roberto Pieraccini, our CEO, gave a talk on the current state of speech recognition and understanding technologies. If you missed it, never fear! We've posted the video of the talk to our YouTube account. A version of this talk was given as a keynote at SpeechTEK, the premier industry conference for speech technologies, on Wednesday, August 21. Check it out below, or visit our YouTube account at www.youtube.com/ICSIatBerkeley. You can also get the slides from this talk as a .pdf.

Here are the details:

Abstract:

Watching Kubrick’s masterpiece “2001 A Space Odyssey,” we realize that we have surpassed most of our technological predictions, except for computers that talk and understand speech like humans do. One of the fundamental reasons for this is that spoken language is so complex, we cannot even satisfactorily explain how it works or how our minds are able to master it. The technology we invented in the 1970s and refined throughout the following 40 years brought the successes of today’s industry. However, speech recognition and understanding systems still work well only in precisely defined domains, and they rapidly degrade when we fall out of optimal conditions. Humans don’t. Even though industry and research have made giant strides, we are still far from the human-like capabilities of our visions and expectations. Where are the missing links? What are we doing wrong? On which problems are researchers around the world spending their cycles? What can we expect in the next few years? This talk, targeted to a heterogeneous audience, aims at answering these questions.

Bio:

Roberto Pieraccini is ICSI's chief executive officer. Prior to joining ICSI, he was the chief technology officer of SpeechCycle, a research manager at IBM T.J. Watson Research and SpeechWorks International, and a member of the technical staff at Bell Labs and AT&T Shannon Laboratories. He started his career in the 1980s as a researcher at CSELT, the research laboratories of the Italian telephone company. His research interests range from speech recognition to spoken language understanding and dialog, multimodal interaction, and machine learning. He is best known for his original contributions to statistical methods for spoken language understanding and machine learning for spoken dialog systems. He is a fellow of the IEEE and International Speech Communication Association, a member of the Applied Voice Input Output Society board, and on the editorial boards of several scientific and technology magazines. His book, The Voice in the Machine, was released in March 2012 by MIT Press.