Skip to Main Content
Porting speech synthesis to mobile platforms is a challenging task. In this paper the application of HMM based speech synthesis on mobile devices is investigated. Several optimization steps (e.g., changing HMM parameters, applying performance specific programming methods) are analyzed in terms of speed, footprint size and MOS scores on three different smartphones. Our goal is to approach real-time functionality with acceptable speech quality. Based on the results optimization steps and resource dependent optimal settings of HMM based speech synthesis are determined.