Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Vocal tract length normalization is an important feature normalization technique that can be used to perform speaker adaptation when very little adaptation data is available. It was shown earlier that VTLN can be applied to statistical speech synthesis and was shown to give additive improvements to CMLLR. This paper presents an EM optimization for estimating more accurate warping factors. The EM formulation helps to embed the feature normalization in the HMM training. This helps in estimating the warping factors more efficiently and enables the use of multiple (appropriate) warping factors for different state clusters of the same speaker.
Mathew Magimai Doss, Julian David Fritsch