MasterVoicing - A whispers to voiced speech assistant

• Autor(a) principal: Maria João Araújo Soutelo
• Data de Publicação: 2017
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://hdl.handle.net/10216/107594
• topic: Engenharia electrotécnica, electrónica e informática; Electrical engineering, Electronic engineering, Information engineering

Aphonia, also known as loss of voice, is a condition that affects the human phonetic system and is characterized by the inability of a speaker to produce normal speech. It can range from partial loss, known as hoarseness, to an almost complete loss of voice, where the voice is nothing but a whisper. Its causes can vary, from physical disease related with injuries, medical procedures or bad habits, such as voice misuse, to mental disorders. Whispering is a natural form of speech for people in some social situations where privacy is desired or silence is recommended. However, for patients with aphonia, whispering is generally their primary way of communication. This can become a problem, because of the difficulty to communicate with other people, and can even cause problems in daily lives or even work related activities. There are some solutions for this problem regarding laryngectomized patients, like the use of an electrolarynx, that recreates an artificial voice, the use of esophageal speech and the tracheo-esophageal puncture with prosthesis, but all of them have some disadvantages and require some degree of practice to master speaking. In terms of technologies, there also exist silent speech interfaces, that are not yet convenient solutions. There are also mobile applications that try to help with this problem, that are generally based in text-to-speech conversion. They require a text input by the user that is followed by its reproduction in speech, resulting in a slow and unnatural usage. Some of these applications function in real-time by a simple click on predefined buttons with text, which also have limitations. With that in mind, the goal in this dissertation is to develop MasterVoicing, a mobile application, for the iOS platform, whose purpose is to give aphonics another alternative to communicate, using their natural way of communicating - whispering. Its validation is verified by the performance of usability tests and its aim is to work in real-time, integrating a whisper-to-speech algorithm that reconstructs natural, voiced, speech from whispers, giving aphonics an easy tool to regain some of their communication freedom, without the drawbacks of other methods that are available to them.