A brain-inspired algorithm improves "cocktail party" listening for individuals with hearing loss

Area of Science:

• Auditory Neuroscience
• Signal Processing
• Hearing Science

Background:

• Selective listening in competing-talker situations is difficult, especially for individuals with hearing loss.
• Existing directional filtering technologies in hearing devices offer modest real-world benefits and often fail in complex acoustic environments.
• Improved speech intelligibility is crucial for communication and daily life participation for those with hearing impairment.

Purpose of the Study:

• To evaluate a novel, biologically inspired algorithm for enhancing speech intelligibility in adults with sensorineural hearing loss.
• To compare the performance of the new algorithm against a standard beamforming approach in a challenging competing-talker task.

Main Methods:

• Development of a biologically inspired algorithm based on a hierarchical network model of the auditory system.
• Utilizing binaural sound inputs to drive neuronal populations tuned to specific spatial locations and frequencies.
• Reconstruction of audible waveforms from the spiking responses of output layer neurons.
• Evaluation of the algorithm with adults experiencing sensorineural hearing loss in a competing-talker scenario.

Main Results:

• The biologically inspired algorithm demonstrated robust intelligibility gains for participants with hearing loss.
• The algorithm succeeded in conditions where a standard beamforming approach proved ineffective.
• Significant improvements in speech understanding were observed, particularly in challenging multi-talker noise.

Conclusions:

• Biologically inspired algorithms show significant potential for aiding individuals with hearing loss in complex listening environments.
• This approach offers a promising advancement over current technologies for overcoming the challenges of "cocktail party" listening.
• The findings support the development of neuro-inspired solutions for hearing assistance devices.