The project draws on long-term research into computer speech processing supported by GACR in the framework of three projects in the years 1996-2007. Their investigation saw unique interconnecting of all leading Czech workplaces dealing with speech recognition and synthesis. Time has now come to use the experience to deepen the knowledge gained and to search for new options and approaches in each area individually. The proposed project deals with computer speech synthesis, bringing together again the prominent Czech workplaces from this area. The main attention will focus on improving the quality of computer-generated speech in the general task of text-to-speech synthesis using new advanced methods of modeling prosodic and signal properties of speech, processing, searching, comparing and selecting sequences of speech units from large inventories of Czech utterances, etc., all with regard to the detailed phonetic knowledge of Czech. The research into the synthesis of emotional speech will also be commenced. All the research involved will be aimed at the synthesis of the Czech language. (en)
zvýšení kvality syntetizované řeči; detekce speciálních fonetických jednotek v hlasových inventářích; tvorba a optimalizace hlasových inventářů; zdokonalení systémů korpusově orientované syntézy řeči; zdokonalení modelů generování prozodie; zahájení výzkumu syntézy české emocionální řeči (cs)