Text-to-speech audio description: audio describing in Italian four short films for the Locarno Shorts Weeks using TTS and SSML

The aim of this thesis is to propose the use of text-to-speech (TTS) technology for recording audio descriptions (AD) in the Italian language, using the Speech Synthesis Markup Language (SSML) to assist the recording process. This process was tested in the production of audio descriptions for four short films for the Locarno Shorts Weeks event, in collaboration with Locarno Film Festival, using Microsoft Azure Neural Text-to-Speech synthetic voices. Additionally, a questionnaire was made to evaluate user perception of TTS AD in Italian. The main objective is to propose an alternative solution to traditional human AD to reduce AD production costs and time, thus enabling an increase in the supply of audio described products. The thesis is divided into five chapters. The first chapter introduces the topic of accessibility, explaining how it relates to Translation Studies and Audiovisual translation. The second chapter dives in the Audiovisual translation modality of audio description, with a focus on legislations, guidelines, and on the modern technologies used to produce it and research on it. The third chapter analyzes one of said technologies, namely text-to-speech. A thorough description of the project and of the applied methodology is given in Chapter 4, while results are analyzed and discussed in Chapter 5.