Robust Detection of Keywords Using Contextual Information (13320N)
IDA Technology Roadmap 2012
This technology falls in the following categories of Singapore's IDA Infocomm Technology Roadmap 2012:
For more information on this technology contact:
This is a method for automatic detection of keywords in speech. The invention can more accurately detect keywords using the contextual information of a word thus reducing misdetection of words that sound similar to the keywords of interest.
The invention comprises a search graph composer and a keyword spotter. Given a set of keywords, the search graph composer combines a word-based and a phoneme-based n-gram language model and generates a hybrid keyword detection graph that consists of the keywords, context words and fillers. The keyword spotter computes the acoustic scores for the paths in the search graph using an acoustic model and uses the combination of the acoustic and language model scores to find the best path. The incorporation of the language model scores around the keywords allows the keyword spotter to more accurately disambiguate the keywords from similar sounding words.
Technology Readiness Level 4 on the scale by the Ministry of Defence Singapore.
About the Research Group
Dr. Khe Chai Sim is an Assistant Professor at the National University of Singapore (NUS). His research focuses on automatic speech recognition, acoustic keyword spotting and other areas of speech processing.