Feld, MBarnard, EVan Heerden, CMuller, C2012-01-182012-01-182009-12Feld, M, Barnard, E, Van Heerden, C and Muller, C. 2009. Multilingual speaker age recognition: regression analyses on the Lwazi corpus. 2009 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU-09), Merano, Italy, 13-17 December 2009978-1-4244-5479-2http://hdl.handle.net/10204/55062009 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU-09), Merano, Italy, 13-17 December 2009Multilinguality represents an area of significant opportunities for automatic speech-processing systems: whereas multilingual societies are commonplace, the majority of speechprocessing systems are developed with a single language in mind. As a step towards improved understanding of multilingual speech processing, the current contribution investigates how an important para-linguistic aspect of speech, namely speaker age, depends on the language spoken. In particular, the authors study how certain speech features affect the performance of an age recognition system for different South African languages in the Lwazi corpus. By optimizing our feature set and performing language-specific tuning, we are working towards true multilingual classifiers. As they are closely related, ASR and dialog systems are likely to benefit from an improved classification of the speaker. In a comprehensive corpus analysis on long-term features, we have identified features that exhibit characteristic behaviors for particular languages. In a follow-up regression experiment, we confirm the suitability of our feature selection for age recognition and present cross-language error rates. The mean absolute error ranges between 7.7 and 12.8 years for same-language predictors and rises to 14.5 years for cross-language predictors.enAutomatic speech recognition systemASRLwazi ASRLwazi corpusGermanic languagesBantu languagesAutomatic speech-processing systemsSpeaker ageSouth AfricaMultilingual corpusSpeaker classificationArticleFeld, M., Barnard, E., Van Heerden, C., & Muller, C. (2009). Multilingual speaker age recognition: regression analyses on the Lwazi corpus. http://hdl.handle.net/10204/5506Feld, M, E Barnard, C Van Heerden, and C Muller "Multilingual speaker age recognition: regression analyses on the Lwazi corpus." (2009) http://hdl.handle.net/10204/5506Feld M, Barnard E, Van Heerden C, Muller C. Multilingual speaker age recognition: regression analyses on the Lwazi corpus. 2009; http://hdl.handle.net/10204/5506.TY - Article AU - Feld, M AU - Barnard, E AU - Van Heerden, C AU - Muller, C AB - Multilinguality represents an area of significant opportunities for automatic speech-processing systems: whereas multilingual societies are commonplace, the majority of speechprocessing systems are developed with a single language in mind. As a step towards improved understanding of multilingual speech processing, the current contribution investigates how an important para-linguistic aspect of speech, namely speaker age, depends on the language spoken. In particular, the authors study how certain speech features affect the performance of an age recognition system for different South African languages in the Lwazi corpus. By optimizing our feature set and performing language-specific tuning, we are working towards true multilingual classifiers. As they are closely related, ASR and dialog systems are likely to benefit from an improved classification of the speaker. In a comprehensive corpus analysis on long-term features, we have identified features that exhibit characteristic behaviors for particular languages. In a follow-up regression experiment, we confirm the suitability of our feature selection for age recognition and present cross-language error rates. The mean absolute error ranges between 7.7 and 12.8 years for same-language predictors and rises to 14.5 years for cross-language predictors. DA - 2009-12 DB - ResearchSpace DP - CSIR KW - Automatic speech recognition system KW - ASR KW - Lwazi ASR KW - Lwazi corpus KW - Germanic languages KW - Bantu languages KW - Automatic speech-processing systems KW - Speaker age KW - South Africa KW - Multilingual corpus KW - Speaker classification LK - https://researchspace.csir.co.za PY - 2009 SM - 978-1-4244-5479-2 T1 - Multilingual speaker age recognition: regression analyses on the Lwazi corpus TI - Multilingual speaker age recognition: regression analyses on the Lwazi corpus UR - http://hdl.handle.net/10204/5506 ER -