REPORT Liang_Idiap-RR-05-2010/IDIAP A Comparison of Supervised and Unsupervised Cross-Lingual Speaker Adaptation Approaches for HMM-Based Speech Synthesis Liang, Hui Dines, John Saheer, Lakshmi EXTERNAL http://publications.idiap.ch/attachments/reports/2010/Liang_Idiap-RR-05-2010.pdf PUBLIC http://publications.idiap.ch/index.php/publications/showcite/Liang_ICASSP_2010 Related documents Idiap-RR-05-2010 2010 Idiap February 2010 The EMIME project aims to build a personalized speech-to-speech translator, such that spoken input of a user in one language is used to produce spoken output that still sounds like the user's voice however in another language. This distinctiveness makes unsupervised cross-lingual speaker adaptation one key to the project's success. So far, research has been conducted into unsupervised and cross-lingual cases separately by means of decision tree marginalization and HMM state mapping respectively. In this paper we combine the two techniques to perform unsupervised cross-lingual speaker adaptation. The performance of eight speaker adaptation systems (supervised vs. unsupervised, intra-lingual vs. cross-lingual) are compared using objective and subjective evaluations. Experimental results show the performance of unsupervised cross-lingual speaker adaptation is comparable to that of the supervised case in terms of spectrum adaptation in the EMIME scenario, even though automatically obtained transcriptions have a very high phoneme error rate. CONF Liang_ICASSP_2010/IDIAP A Comparison of Supervised and Unsupervised Cross-Lingual Speaker Adaptation Approaches for HMM-Based Speech Synthesis Liang, Hui Dines, John Saheer, Lakshmi decision tree marginalization HMM state mapping unsupervised cross-lingual speaker adaptation EXTERNAL http://publications.idiap.ch/attachments/papers/2009/Liang_ICASSP_2010.pdf PUBLIC http://publications.idiap.ch/index.php/publications/showcite/Liang_Idiap-RR-05-2010 Related documents Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing Dallas, U.S.A. 2010 March 2010 4598-4601 The EMIME project aims to build a personalized speech-to-speech translator, such that spoken input of a user in one language is used to produce spoken output that still sounds like the user's voice however in another language. This distinctiveness makes unsupervised cross-lingual speaker adaptation one key to the project's success. So far, research has been conducted into unsupervised and cross-lingual cases separately by means of decision tree marginalization and HMM state mapping respectively. In this paper we combine the two techniques to perform unsupervised cross-lingual speaker adaptation. The performance of eight speaker adaptation systems (supervised vs. unsupervised, intra-lingual vs. cross-lingual) are compared using objective and subjective evaluations. Experimental results show the performance of unsupervised cross-lingual speaker adaptation is comparable to that of the supervised case in terms of spectrum adaptation in the EMIME scenario, even though automatically obtained transcriptions have a very high phoneme error rate.