A Unit Selection based Synthesis (USS) system selects appropriate units from the inventory based on target cost and join cost. For a syllable based USS system, it is not practically feasible to cover all possible syllables in a language with all possible contexts. Hence, some discontinuity is observed at the concatenation points in the synthesized speech due to unavailability of units with proper target specifications in the inventory. In this work, we are using prosody modification of the syllables before concatenation to match fundamental frequency (F0) and intensity of adjacent syllables. This approach is also helpful to build a small footprint syllable level USS system, where many examples of the basic units with varying prosodical and spectral features is not available. The discontinuity in pitch and intensity at the concatenation point is reduced after prosody modification. Moreover, naturalness and intelligibility of synthesized speech is found to be improved using the proposed method. © 2015 IEEE.