This technology is a method that uses neural signals to selectively separate and enhance auditory signals from particular speakers in a noisy environment.
Hearing loss, affecting 360 million people globally, poses substantial challenges in social interactions. The primary issue with current hearing aids is their inability to effectively isolate and amplify individual speakers' voices in noisy settings. Addressing this shortcoming can greatly improve overall quality of life and ease daily interactions in different auditory environments.
This Brain-Informed Speech Separation (BISS) system discerns auditory focus using neural data from subjects, captured through either invasive intracranial electroencephalography (iEEG) or non-invasive EEG. BISS applies the neural patterns corresponding to speech attention to a sophisticated deep learning algorithm for precise, unbiased extraction of the targeted auditory signal. The specificity of auditory extraction in this system is indicative of its potential in enhancing speech clarity in noisy environments. The efficacy of BISS in isolating the attended speech, even amidst multiple sound sources, shows promise for developing advanced neuro-steered hearing-assistive devices.
Patent Issued (US 11875813)