After considerably long time (and with help of Davide Iemmola, my friend from CSM–MACD) I managed to code Italian alphabet. This posed rather different difficulties than in case of Polish alphabet, as instead of compressing the alphabet to fit my set of 28 variables (suited perfectly for English!) I had to extend the content. I managed to do this in a truly surgical fashion by throwing 5 noises away. My decision was to cut on the double-noise pairs, so the only doubles to remain would be 'space' and 'start' which remain constant in each language coding.
As said before, Italian alphabet consists of 21 letters (ABCDEFGHILMNOPQRSTUVZ), 10 digraphs (ch, ci, gh, gi, gl, gn, ha, qu, sc, zz) and 3 trigraphs (gli, sch, sci). These rules helped with assigning right letters to right noises.