CONF Saheer_ICASSP_2012/IDIAP COMBINING VOCAL TRACT LENGTH NORMALIZATION WITH HIERARCHIAL LINEAR TRANSFORMATIONS Saheer, Lakshmi Yamagishi, Junichi Garner, Philip N. Dines, John constrained structural maximum a posteriori linear regression hidden Markov models speaker adaptation Statistical parametric speech synthesis vocal tract length normalization EXTERNAL https://publications.idiap.ch/attachments/papers/2012/Saheer_ICASSP_2012.pdf PUBLIC https://publications.idiap.ch/index.php/publications/showcite/Saheer_Idiap-RR-11-2012 Related documents Proceedings in International conference on Speech and Signal processing Kyoto, Japan 2012 IEEE SPS (ICASSP) 4493-4496 Recent research has demonstrated the effectiveness of vocal tract length normalization (VTLN) as a rapid adaptation technique for statistical parametric speech synthesis. VTLN produces speech with naturalness preferable to that of MLLR based adaptation techniques, being much closer in quality to that generated by the original average voice model. However with only a single parameter, VTLN captures very few speaker specific characteristics when compared to linear transform based adaptation techniques. This paper proposes that the merits of VTLN can be combined with those of linear transform based adaptation in a hierarchial Bayesian framework, where VTLN is used as the prior information. A novel technique for propagating the gender information from the VTLN prior through constrained structural maximum aposteriori linear regression (CSMAPLR) adaptation is presented. Experiments show that the resulting transformation has improved speech quality with better naturalness, intelligibility and improved speaker similarity. REPORT Saheer_Idiap-RR-11-2012/IDIAP Combining Vocal Tract Length Normalization with Linear Transformations in a Bayesian Framework Saheer, Lakshmi Yamagishi, Junichi Garner, Philip N. Dines, John EXTERNAL https://publications.idiap.ch/attachments/reports/2011/Saheer_Idiap-RR-11-2012.pdf PUBLIC Idiap-RR-11-2012 2012 Idiap April 2012 Recent research has demonstrated the effectiveness of vocal tract length normalization (VTLN) as a rapid adaptation technique for statistical parametric speech synthesis. VTLN produces speech with naturalness preferable to that of MLLR- based adaptation techniques, being much closer in quality to that generated by the original average voice model. By contrast, with just a single parameter, VTLN captures very few speaker specific characteristics when compared to the available linear transform based adaptation techniques. This paper proposes that the merits of VTLN can be combined with those of linear transform based adaptation technique in a Bayesian framework, where VTLN is used as the prior information. A novel technique of propa- gating the gender information from the VTLN prior through constrained structural maximum a posteriori linear regression (CSMAPLR) adaptation is presented. Experiments show that the resulting transformation has improved speech quality with better naturalness, intelligibility and improved speaker similarity.