Text Normalization and Phonetic Analysis Modules for Macedonian TTS synthesis

The paper presents the Text Normalization and Phonetic Analysis Modules that are part of the frontend of the text-to-speech (TTS) system “Speak Macedonian”. First, the architecture of the frontend of the TTS system “Speak Macedonian” is shortly presented, followed by a detailed look into the two modules. For each of the modules a short summary is given of the tasks and developed solutions, found in literature. In following, the tasks specific to the orthography and orthoepy of the standard Macedonian language, as well as the designed algorithms, are presented in some detail. At the end, the results are shortly discussed and deemed as adequate for the task at hand, and directions for further improvement are pointed out.