2nd Workshop on Spoken Language Technologies for Under-Resourced Languages

Universiti Sains, Penang, Malaysia
May 3-5, 2010

Adaptation Techniques for Speech Synthesis in Under-Resourced Languages

Gopala Krishna Anumanchipalli, Alan W. Black

Language Technologies Institute, Carnegie Mellon University, Pittsburgh, PA, USA

This paper presents techniques for building speech synthesizers targeted at limited data scenarios - limited data from a target speaker; limited or no data in a target language. A resource sharing strategy within speakers and languages is presented giving promising directions for under-resourced languages. Our results show the importance of the amount of training data, the selection of languages and the mappings across languages in a multilingual setting. The objective evaluations conclusively prove that the presented adaptation techniques are well suited for building voices in resource-scarce conditions.

Index Terms: Speech Synthesis, Adaptation, Voice conversion, under-resourced languages.

Full Paper

Bibliographic reference.  Anumanchipalli, Gopala Krishna / Black, Alan W. (2010): "Adaptation techniques for speech synthesis in under-resourced languages", In SLTU-2010, 51-55.