Main Profile

At A Glance

Speech recognition and retrieval using unsupervised sub-word language models

Google Tech TalksFebruary, 8 2008ABSTRACTUnsupervised morpheme analysis and language models developed atHelsinki University of Technology (TKK) open interesting new views onlarge vocabulary speech recognition, information retrieval and machinetranslation.This approach learns suitable sub-word units directly from relevanttext corpora in a completely data-driven manner and can, thus, beeasily ported to various morphologically complex languages.Our system includes several public domain software packages: Morfessorfor determining the modeling units, VariKN to train effectivelysmoothed long-span LMs, a near-realtime single-pass decoder for LMsand LVCSR, and demos at: http://www.cis.hut.fi/projects/morpho/http://www.cis.hut.fi/projects/speech/I will also summarize the results from the recent Morpho Challengemachine learning competitions in morpheme analysis for languagemodeling and information retrieval:http://www.cis.hut.fi/morphochallenge2007/Speaker: Mikko KurimoMikko Kurimo is currently Academy Research Fellow at HelsinkiUniversity of Technology (TKK), where he was an acting professor ofComputer and Information Science (2001-2003). In 1998-2000 he worked as a postdoc at IDIAP and his Doctor of Science (PhD) degree is from TKK 1997. He has a long research record and a number of publications in largevocabulary speech recognition (since 1990) and has been involved in severalinternational and national research projects and is TKK's site managerin the FP6 Network of Excellence PASCAL.At TKK's Adaptive Informatics Research Centre (which is one of the CentresofExcellence nominated by the Academy of Finland) he leads the SpeechRecognition and Multimodal Interfaces research groups.The current research focus of his speech group is in language independentand unsupervised models for continuous speech with morpheme-basedlanguage models for very large vocabulary.The relevant pilot applications in the group range from unlimitedvocabulary dictation systems for different languages to audio indexingand speech-to-speech translation.At the moment Mikko Kurimo is an International Fellow at SRI untilFebruary 2008.
Length: 48:40

Contact

Questions about Speech recognition and retrieval using unsupervised sub-word language models

Want more info about Speech recognition and retrieval using unsupervised sub-word language models? Get free advice from education experts and Noodle community members.

  • Answer

Ask a New Question