Development of Speech Syntheses for Lower Sorbian and Upper Sorbian using MaryTTS

Lower and Upper Sorbian are Slavic languages spoken in eastern Germany. As minority languages they are phonetically and phonologically less well documented than major languages and so far only benefit from limited digital resources. Thus, they are usually not considered for technologies such as text-to-speech (TTS) synthesis. This paper describes our approach, methods, and resources for creating two TTS synthesis systems for the Sorbian languages using the open-source MaryTTS platform. All data and resources for the unit-selection synthesis voices will be published under an open-source license and will be made freely available.