Creating conversational interfaces for children

摘要: Creating conversational interfaces for children is challenging in several respects. These include acoustic modeling automatic speech recognition (ASR), language and dialog modeling, multimodal-multimedia user interface design. First, issues ASR of children's are introduced by an analysis developmental changes the spectral temporal characteristics signal using data obtained from 456 children, ages five to 18 years. Acoustic adaptation vocal tract normalization algorithms that yielded state-of-the-art performance on described. Second, experiment designed better understand how interact with machines spoken Realistic multimedia interaction were 160 who played a voice-activated computer game Wizard Oz (WoZ) scenario. Results these developing novel models as well unified maximum likelihood framework decoding semantic classification understanding Leveraging lessons learned WoZ study concurrent experience evaluation, personal agent prototype was designed. Details architecture application details Informal evaluation found positive especially animated interface.