Kleynhans, NTBarnard, E2012-02-232012-02-232005-11Kleynhans, NT and Barnard, E. Language dependence in multilingual speaker verification. Sixteenth Annual Symposium of the Pattern Recognition Association of South Africa, Langebaan, South Africa, 23-25 November 20050-7992-2264-Xhttp://www.prasa.org/proceedings/2005/prasa05-18.pdfhttp://hdl.handle.net/10204/5591Sixteenth Annual Symposium of the Pattern Recognition Association of South Africa, Langebaan, South Africa, 23-25 November 2005An investigation into the performance of current speaker verification technology within a multilingual context is presented. Using the Oregon Graduate Institute (OGI) Multi-Language Telephone Speech Corpus (MLTS) database, the authors found that the performance of textindependent speaker verification depends fairly strongly on the language being spoken, with equal error rates differing by more than a factor of three between the best and worst performing languages. It was also found that training language-specific universal background models, to normalize speakers' scores, gives better results than both language-independent background models and background models derived from relevant language families.enSpeaker verificationUBMsLanguage differencesOGI MLTSConference PresentationKleynhans, N., & Barnard, E. (2005). Language dependence in multilingual speaker verification. PRASA. http://hdl.handle.net/10204/5591Kleynhans, NT, and E Barnard. "Language dependence in multilingual speaker verification." (2005): http://hdl.handle.net/10204/5591Kleynhans N, Barnard E, Language dependence in multilingual speaker verification; PRASA; 2005. http://hdl.handle.net/10204/5591 .TY - Conference Presentation AU - Kleynhans, NT AU - Barnard, E AB - An investigation into the performance of current speaker verification technology within a multilingual context is presented. Using the Oregon Graduate Institute (OGI) Multi-Language Telephone Speech Corpus (MLTS) database, the authors found that the performance of textindependent speaker verification depends fairly strongly on the language being spoken, with equal error rates differing by more than a factor of three between the best and worst performing languages. It was also found that training language-specific universal background models, to normalize speakers' scores, gives better results than both language-independent background models and background models derived from relevant language families. DA - 2005-11 DB - ResearchSpace DP - CSIR KW - Speaker verification KW - UBMs KW - Language differences KW - OGI MLTS LK - https://researchspace.csir.co.za PY - 2005 SM - 0-7992-2264-X T1 - Language dependence in multilingual speaker verification TI - Language dependence in multilingual speaker verification UR - http://hdl.handle.net/10204/5591 ER -