This report attempts to determine whether improving the phonetic match between a lexicon used in speech synthesis and the speech of the speaker who provides the source for a synthetic voice improves the quality and authenticity of the synthetic voice. In order to explore this question, a survey is made of the literature describing the accent of urban Glasgow, and these data serve as the basis for the accent’s implementation in the Unisyn accent-independent lexicon system. After implementation, two voices are built from the accent, one using an automatic labelling procedure and one using a corrected labelling. The voices are assessed subjectively for quality and compared in forced-choice listening tests to reference voices built using the Edinburgh accent of the Unisyn lexicon. The results are inconclusive, but the method is generalizable and potentially powerful.