Text-to-Speech Synthesis for Mobile Phones

MOBITALK (CP-ISL-80) aims at the integration of speech technologies into mobile networks and devices. Coordinated research and development efforts towards this aim included the creation of new and, the improvement of, existing speech synthesis algorithms and practices, the implementation and setup of new applications in mobile phones, the development of applications for the short messaging centres of mobile networks as well as research in user requirements and the commercial exploitation of project results. MOBITALK addresses visually impaired individuals incapable of using contemporary mobile phones. It also addresses those who incidentally have difficulty using their devices, for example while driving. In order to facilitate service use irrespective of the capabilities of terminal devices application development was performed on two complementary levels: (a) at the mobile device and (b) at the mobile network. These two levels impose different requirements both computational and in terms of quality of synthetic speech. Mobile devices have limited capabilities thereby imposing substantial constraints both in memory and computations. For the transfer of speech synthesis in such devices it is therefore imperative the reduction of the computational requirements without however compromising the quality of the synthetic speech. On the contrary at the network level, the computational capabilities are greater permitting the installation and configuration of high quality TtS systems.