In this work we present an intelligibility evaluation experiment for Italian synthesizers that uses an open-response test in which listeners are instructed to simply write down what they heard on each trial. Six systems and a natural voice as a reference have been evalutated both over the orthophonic and telephonic channels. 57 VCV and 51 CV are used. This material allows the evaluation of consonants in both initial and central position. A dedicated software called SOAP (Speech Output Assessment Package) is used.
Keywords: text-to-speech synthesizers intelligibility evaluation, adverse conditions