Portal de Periódicos da CAPES

Real-time speech synthesis for neural prosthesis.

2009; Acoustical Society of America; Volume: 125; Issue: 4_Supplement Linguagem: Inglês

10.1121/1.4783342

1520-9024

Frank H. Guenther,

Robotics and Automated Systems

The idea of building an artificial vocal tract that could serve as a prosthetic device for mute individuals has been pursued since the pioneering work of Wolfgang von Kempelen and contemporaries in the 18th century. These early speaking machines were built out of wood, metal, rubber, and other materials fashioned into a mechanical vocal tract whose shape was manipulated manually by the user while air was forced through it. In the mid-20th century, Gunnar Fant and colleagues constructed electrical resonating circuits capable of producing speechlike sounds over loudspeakers, and shortly thereafter fully computer-based speech synthesizers were developed, including formant synthesizers and articulatory synthesizers. The recent development of brain-computer interfaces driven by neuronal activity raises the possibility of a speech prosthesis that would allow mute paralyzed individuals to directly control a speech synthesizer to create speech output. We have implemented such a brain-computer interface in a 25-year-old volunteer suffering from locked-in syndrome, characterized by complete loss of movement but intact perception and consciousness. To date the volunteer has successfully used the prosthesis to produce vowel sounds. Future developments in speech synthesis necessary for allowing full conversational speech via brain-computer interface will be outlined. [Work supported by NIH/NIDCD and NSF.]