Silent Speech Interface Using Manifold Decoding Of Biosignals

Abstract

Researchers at the University of California, Davis have developed a technology that provides a novel method for decoding biosignals into speech, enhancing communication for individuals with speech impairments.

Full Description

The technology involves a computer-implemented method and system for decoding biosignals (e.g., those indicative of orofacial movements) into speech. It utilizes a unique approach that reduces the computational complexity, and thus the amount of time needed, to decode biosignals and translate them into synthesized speech.

Applications

• Assistive technologies for individuals with speech impairments due to ALS, stroke, cancer, and other conditions. 
• Human-computer interaction systems that require robust speech recognition capabilities. 
• Medical devices and applications focused on rehabilitation and communication restoration. 
• Can be used to decode/translate a wide variety of biosignals that are recorded from patients.

Features/Benefits

• Addresses the variability of biosignals across individuals and sessions, enhancing accuracy and robustness. 
• Reduces the computational demand and need for extensive retraining typically associated with neural network-based approaches. 
• Improves accessibility for individuals with speech impairments due to various causes, including neurological diseases and physical damage. 
• Facilitates real-time communication by efficiently decoding complex biosignals into speech. 
• Overcomes communication barriers faced by individuals with dysarthria, dysphonia/aphonia, and other speech impairments. 
• Addresses the challenge of signal variability due to individual anatomical and physiological differences. 
• Reduces the high computational cost and inefficiency of existing neural network approaches in adapting to new individuals.