Voice Gestures (VG) is an application that gives AAC users the ability to say what they want to say in a way they choose to say it. VG gives AAC users the means to control how their voice sounds to others, thereby establishing themselves as people with distinct personalities as perceived through their speech, in the manner of people with normal vocal capabilities. VG uses an innovative approach to detect icons in a text string and to translate them into control parameters for a formant-type synthesizer, such as the one used in the DECtalk Text-To-Speech system. A user enters icons along with text or symbols into an AAC device and the VG tool translates each icon into synthesizer parameter strings that control the acoustic characteristics of the synthesized speech. VG is aimed at an underserved population - individuals who cannot use their biological voices to communicate easily with family, friends and their community, and instead rely on using an AAC device that generates synthetic speech. In devices that are currently available, users have very limited control of the way their message sounds. Thus, misunderstandings result and the AAC user becomes frustrated. Another reason indicating the need for the VG tool is that having a way to communicate more clearly will facilitate conversations between users of AAC devices and the others in their worlds, and thus increase their social interactions. Breakdown of communication between partners in an AAC device-mediated communication has been frequently noted. As a result of these frequent breakdowns, users of AAC devices report that many persons are reluctant to spend time interacting with them. We will design a set of VG icons that give device users control of tailoring their own voices to their intentions. Based on recommendations of actual users of AAC devices, we will develop 5 voice gestures (contrastive emphasis, loud and clear reply, tentative answer, sarcasm, and fast reply), synthesize 15 short phrases or words with each voice gesture, and evaluate the ease of use of the icons and the appropriateness of the synthesis. These varieties of voice gesture have been reported by users of AAC devices to be needed. Subjects for testing and evaluations will be both users of AAC devices and their interlocutors. Voice Gestures (VG) add expressiveness to synthetic speech for individuals who cannot use their biological voices to communicate easily with family, friends and their community. Users enter icons along with text or symbols into an AAC device, and the VG tool translates each icon into synthesizer parameter strings that control the acoustic characteristics of the synthesized speech. ? ? ?
