US researchers claim their new hearing aid algorithm could improve word recognition accuracy in noisy situations by up to 40%.
The Boston University (BU) researchers said the brain-inspired algorithm for sound processing could help hearing aids tune out interference and isolate single talkers in a crowd of voices.
In testing done in 2023, they found it improved word recognition accuracy by 40% relative to current hearing aid algorithms, the researchers wrote in Communications Engineering on 22 April 2025. However, no hearing aid devices were used in the experiment so they do not make any performance claims against hearing aids on the market, the lead author told HPA.
The algorithm’s developer, Associate Professor Kamal Sen, has patented the ‘biologically oriented sound segregation algorithm’ or BOSSA, and wants to connect with companies interested in licensing the technology.
“We were extremely surprised and excited by the magnitude of the improvement in performance – it’s pretty rare to find such big improvements,” he said on the university’s news website, The Brink.
The researchers evaluated the algorithm in adults with sensorineural hearing loss, using a challenging competing-talker task. The algorithm “led to robust intelligibility gains under conditions in which a standard beamforming approach failed,” they said.
“The results provide compelling support for the potential benefits of biologically inspired algorithms for assisting individuals with hearing loss in “cocktail party” situations,” they said.
The “cocktail party problem” refers to the difficulty people have hearing, especially those with hearing loss, in noisy situations such as a party, restaurant or bar.
A/Prof Sen said the algorithm’s main aim was to use spatial cues to perform sound segregation and selection, not sound localisation.
A physicist who later trained in neuroscience, A/Prof Sen is a BU College of Engineering associate professor of biomedical engineering and director of the university’s Natural Sounds and Neural Coding Laboratory which is affiliated with BU’s centres for neurophotonics and systems neuroscience.
Australian link to NAL
A co-author, former research scientist at Australia’s National Acoustic Laboratories, Associate Professor Virginia Best, is now a Boston University Sargent College of Health & Rehabilitation Sciences research associate professor of speech, language, and hearing sciences.
“The primary complaint of people with hearing loss is that they have trouble communicating in noisy environments,” she said. “These environments are common in daily life and tend to be really important to people so solutions that can enhance communication in noisy places have the potential for a huge impact.”
A/Prof Best, A/Prof Sen and biomedical engineering PhD candidate Mr Alexander Boyd from Boston University’s Hearing Research Center, Department of Biomedical Engineering, and Department of Speech, Language and Hearing Sciences, tested the ability of current hearing aid algorithms to cope with the noise of cocktail parties.
They said many hearing aids included noise reduction algorithms and directional microphones, or beamformers to emphasise sounds coming from the front. They benchmarked against the industry standard algorithm in hearing aids in 2023.
For the past 20 years, A/Prof Sen has studied how the brain encodes and decodes sounds, looking for circuits involved in managing the cocktail party effect. With researchers in his laboratory, he plotted how sound waves are processed at different stages of the auditory pathway, tracking their journey from the ear to translation by the brain.
A key mechanism is that inhibitory neurons are brain cells that help suppress certain, unwanted sounds, he said, a form of internal noise cancellation.
“If there’s a sound at a particular location, these inhibitory neurons get activated,” he said, adding that different neurons were tuned to different locations and frequencies.
The brain’s approach was the inspiration for the new algorithm, he said. It used spatial cues such as volume and timing of a sound to tune into or out of it, sharpening or muffling a speaker’s words as needed.
“It’s basically a computational model that mimics what the brain does, and it segregates sound sources based on sound input,” he said.
“The algorithm is based on a hierarchical network model of the auditory system, in which binaural sound inputs drive populations of neurons tuned to specific spatial locations and frequencies, and the spiking responses of neurons in the output layer are reconstructed into audible waveforms,” they wrote.
A/Prof Best said the only way to know if a benefit would translate to the listener was through behavioural studies. An expert on spatial perception and hearing loss, she helped design a study using young adults with sensorineural hearing loss, typically caused by genetic factors or childhood diseases.
Participants wore headphones that simulated people talking from different nearby locations. Their ability to pick out select speakers was tested using the new algorithm, the current standard algorithm, and no algorithm.
Boyd who helped collect much of the data and was the lead author on the paper, said the standard beamforming algorithm benchmarked against was binaural MVDR (Minimum Variance Distortionless Response), the implementation of which was provided by researchers at Demant.
“One limitation we placed on MVDR in this experiment was only using two binaural microphones instead of the four microphones often used on hearing aids,” he told Hearing Practitioner Australia.
“This was to make the comparison fair but MVDR in hearing aids that uses two mics may perform better than the version tested, and we are currently working on new real-time comparisons to four mic MVDR.
“None of the stimuli or processing in this experiment was performed in a hearing aid; rather the audio were pre-preprocessed offline, so we cannot say the comparison was against any particular device on the market but rather the underlying beamforming used in most hearing aids.”
Boyd said no hearing aid devices were used in the experiment. “Our only claim is that in these experimental conditions, we saw intelligibility boosts of ~40% between binaural MVDR processed audio and BOSSA processed.
“We are beginning experiments to compare real-time implementations of BOSSA vs hearing aids in these kinds of conditions, but for now we do not make any performance claims against on the market hearing aids.”
They’re also testing an upgraded version that incorporates eye tracking technology to allow users to better direct listening attention. They said the science powering the algorithm might also have implications beyond hearing loss to do with attention in ADHD or autism.
