Systém pro syntézu mluvených hlásek pro tvorbu umělého mluveného signálu

Abstract

The aim of this Master's thesis is to design and implement a concatenative speech synthesizer in Czech language. This system is built on linear predictive coding based vocal tract model and converts written text to a synthetic speech signal. Units are segmented from previously recorded prompts, encoded by algorithms and stored in inventories. The resulting speech is created by concatenation of reconstructed segments by one of the three selected methods. Lastly, the synthesizer is evaluated by MRT and MOS tests.