Add new comment

The archives from 1993 provide a wonderful example of ICSI's twin traditions of first-rate speech research and creative acronyms: the Berkeley Restaurant Project, or BeRP. The Realization Group, which would later be renamed the Speech Group, developed a spoken dialog system that gave restaurant recommendations based on a user's spoken responses to questions asked by the system.

The system's audio recognition used the newly developed relative spectral processing technique (RASTA), which helps machines handle changes in the audio spectrum. At the time, most speech systems had difficulty, for example, dealing with audio recorded on different microphones from those used to record its training data. RASTA, by comparison, was more sensitive to spectral changes in the audio over time and less sensitive to the spectrum itself. This kind of processing helps systems handle the sometimes drastic differences in spectrum between the data used for training the models and the data used for testing them. With RASTA, what’s important is the change from one moment to the next, not the absolute audio spectrum at any given point. This became particularly important later, when cell phones were ubiquitous and RASTA was designed into millions of phones.

In addition to RASTA, BeRP used a natural language backend that parsed words and produced database queries, a restaurant database, and accent detection and modeling algorithms that helped the system understand foreign accents and different pronunciations.

The system was unusual in that both the system and its users could initiate questions, and the system could continue a conversation even when users did not respond directly to its questions. The system was developed by group leader Professor Nelson Morgan, along with postdoctoral fellows Gary Tajchmann and Dan Jurafsky, and graduate students Chuck Wooters and Jonathan Segal.

Chuck, who was Morgan’s first graduate student, said the system had a tight integration between natural language understanding and speech recognition. “You didn’t think of the speech recognizer as static,” he said. “It was more of a living system.”

Read more about the Speech Group's accomplishments.