VocaliD: Personal voices for augmented communicators

Abstract

The goal of the VocaliD project (for vocal identity) is to develop personalized synthetic voices for children and adults who rely on speech generating devices (SGDs) for verbal communication. Our approach extracts acoustic properties related to source, vocal tract, or both from a target talker’s disordered speech (whatever sounds they can still produce) and applies these features to a synthetic voice that was created from a surrogate voice donor who (ideally) is similar in age, size, gender, etc. The result is a synthetic voice that contains as much of the vocal identity of the target talker as possible yet the speech clarity of the surrogate talker’s synthetic voice. To date, we have deployed several synthetic voices using this technology. Three case studies will be presented to illustrate the methods used in voice generation and the results from three pediatric SGD users. We will also describe plans to greatly extend our database of surrogate voice donor speech, allowing us to better match regional/dialectical features to the needs of the target SGD users.