In this paper, a segmentation method based on a variant of the Itakura distance is presented together with its implementation in a speech signal segmentation facility. With this facility the quality of speech segments is evaluated at the early stage of acquisition. Thus, the construction of a segment inventory is significantly accelerated. The facility was used for the construction of a segment inventory consisting of more than 400 speech segments. The quality of the segments was informally evaluated by synthesizing more than 1500 words of the Greek language that are widely used in everyday discourse. The work was motivated by the construction of a TTS based on the syllable concatenation approach, aiming at producing an inflectional environment for hundreds of verbs and nouns. The resulting system of the undertaken project will be integrated into an educational platform for language learning for elementary school students.