Mark Edward Epstein

A cloud on the horizon

In this talk, I will present an overview of many areas in speech and language processing that Google is researching, developing and deploying to users. As the amount of information on the web grows, difficult problems in speech, text, information retrieval and natural language processing are now being solved and new challenges are appearing. For example, Google now builds speech recognition language models for a one million word vocabulary from billions of n-grams. By using data-driven approaches, Google is improving its ability to deliver information to global users of computers, tablets and smartphones.