A virtual vocal tract for speech output using non-invasive brain-computer interface

点击翻译按钮获取中文摘要

Brain-computer interfaces (BCIs) have the potential to aid the restoration and rehabilitation of movement and communication for individuals with profound neuromotor impairments, including loss of speech. The current therapeutic paradigm for communication BCIs focuses on typing or spelling interfaces, which satisfy an immediate need for those who are not able to use other assistive technology for communication. Many users of BCI; however, desire an additional option to restore fluent vocal production beyond the capabilities of current BCI spellers. An alternative, complementary approach to communication BCIs envisions their potential as a replacement for the human vocal tract. The scientific foundation for a brain-controlled virtual vocal tract has been building in recent years from neuroscience studies of speech production, development of silent speech devices, and improvements to speech synthesizer technology. Our recent study proved the feasibility of a BCI-controlled speech synthesizer, though it was limited to the production of vowel sounds. There is now sufficient evidence and technology to bridge the gap between these past approaches toward a more complete BCI device that can continuously produce both vowels and consonants. The production of both types of phonemes is critical for building consonant-vowel pairs (e.g., syllables), words, and sentences through practice in a process similar speech development in young children. Importantly, the use of a replacement virtual vocal tract BCI depends on one's ability to learn new speech motor skills, which is not immediately clear for individuals with severe neuromotor impairments with dysarthria or anarthria. The goal of this study is to determine whether individuals with neuromotor impairments can learn new speech motor skills using a brain-controlled virtual vocal tract despite severe speech motor dysfunction. To address this goal, nine adults with severe speech intelligibility deficits due to progressive (e.g., amyotrophic lateral sclerosis) and non-progressive (e.g., brainstem stroke) neuromotor dysfunction, as well as 18 age and sex- matched control participants, will be trained to operate a virtual vocal tract BCI. The aims of the proposed study are: (Aim 1) to determine whether participants are able to learn new speech motor skills for accurate production of speech movements (including syllables) using a BCI virtual vocal tract, and (Aim 2) to use the N100 event-related potential suppression response to determine whether the functional network of brain regions needed for speech motor control is still intact and active, particularly for participants with neuromotor impairments. The results of this research will challenge and update current theories of speech motor control in individuals with neuromotor impairments and improve on the design and implementation of BCIs for controlling speech synthesizers in real- time. Both goals advance the long-term goal of a neural prosthesis for restoration of speech capable of providing continuous auditory output for individuals with severe speech impairment. This long-term goal has specific clinical relevance to individuals with severe neuromotor disorders and / or paralysis (e.g., locked-in syndrome) and has the potential to improve patients' quality of life and allow them to better engage in social interactions.