中央研究院 資訊科學研究所

Recently we have proposed a novel keyword-aware language modeling framework to incorporate side information available from a given list of keywords for LVCSR-based keyword search (KWS) and demonstrated that language modeling with keyword-specific information yields significant gains over conventional n-grams in system performance. We explore the potential of using keyword-aware language modeling to extend the capability of LVCSR-based KWS by trading higher false alarm rates in exchange for lower missed detection rates. We demonstrate empirically that by using keyword-aware language modeling, we are able to achieve a tunable system performance with different operating points. We also present context-depend fillers to reduce false alarms across all operating points while preserving the desirable low miss detection rates. Using only 10 hours on transcribed speech data to train the acoustic and language models and testing on the IARPA Babel 2013 OpenKWS in Vietnamese and 2014 OpenKWS in Tamil, the proposed keyword-aware language modeling improves conventional n-gram by 50% in a commonly-adopted evaluation metric, called actual term weighted value (ATWV).

Chin-Hui Lee is a professor at School of Electrical and Computer Engineering, Georgia Institute of Technology. Before joining academia in 2001, he had 20 years of industrial experience ending in Bell Laboratories, Murray Hill, New Jersey, as a Distinguished Member of Technical Staff and Director of the Dialogue Systems Research Department. Dr. Lee is a Fellow of the IEEE and a Fellow of ISCA. He has published over 400 papers and 30 patents, and was highly cited for his original contributions with an h-index of 66. He received numerous awards, including the Bell Labs President's Gold Award in 1998. He won the SPS's 2006 Technical Achievement Award for "Exceptional Contributions to the Field of Automatic Speech Recognition". In 2012 he was invited by ICASSP to give a plenary talk on the future of speech recognition. In the same year he was awarded the ISCA Medal in scientific achievement for “pioneering and seminal contributions to the principles and practice of automatic speech and speaker recognition”.