details
The silent speech concept, when applied to Human-Computer Interaction (HCI), describes a system wich allows for speech communication in the absence of an acoustic signal.
The silent speech concept, when applied to Human-Computer Interaction (HCI), describes a system which allows for speech communication in the absence of an acoustic signal. Silent Speech Interfaces(SSI) appear as an alternative or complementary approach that provides Automatic Speech Recognition (ASR) for HCI in many contexts: in harsh environments; where speaking at an audible rate may disturb others or disclose confidential information; to handle elderly or children speech; and, more importantly, it provides a natural interface for speech-impaired users. Despite recent evolution, performance and usability of SSI systems still have much room for improvement. An increased performance of such systems would enable their application in relevant areas, such as Ambient Assisted Living, leveraging their potential. In this context we have established several goals for this project: (1) Perform European Portuguese (EP) adoption of the SSI; (2) Detect EP nasality with the multimodal SSI, by overcoming identified limitations of current techniques (3) develop a multimodal HCI approach for SSI; and (4) explore the use of supplementary direct measures acquired from more invasive modalities in multimodal SSI setups, to be used as ground truth in articulation processes, enhancing our comprehension of less invasive modalities.