An articulatory synthesiser has been developed that uses the Kelly-Lochbaum filter structure. The vocal tract area function is generated from nine 'quasi-articulatory' parameters. Sets of parameters obtained for several vowels by magnetic resonance imaging have been used to mark the ends of vowel-vowel transitions. A codebook has been obtained for several transitions using an inching technique designed to ensure a uniform coverage in formant space, and that smooth changes in area function occur along straight formant tracks. A non-linear regression technique has then be used to obtain an approximate cubic relationship between formant frequencies and synthesiser parameters. Synthesis of several target tracks have produced synthetic tracks that are similar.