This paper presents a speech synthesizer (SS) system developed for the Marathi language, specifically targeting sentences from the primary education domain. Due to the inherently long sentences in this domain, they are processed and combined using the SS engine, which converts text input into spoken words. The system is implemented using a unit-selection approach based on the uniqueness of speech units, ensuring natural-sounding output. The methodology focuses on capturing the natural characteristics of Marathi speech. To assess the quality of artificial voice generation, a prosody analysis was conducted. This involved measuring the intensity and duration of generated voice signals, with their intensities quantified through mean and standard deviation calculations. Additionally, a subjective listening test was performed to evaluate the intelligibility and naturalness of the machine-generated voice. The results demonstrate that the developed Marathi SS achieved satisfactory performance. Overall, this study highlights the effectiveness of a unit-selection approach for Marathi speech synthesis, providing a strong foundation for further advancements in regional language speech technology.