AI in Hearing Aids

Every modern digital hearing aid is a tiny computer. It picks up sound through microphones, splits it into frequency bands, amplifies each band to match the wearer's hearing loss, and plays it through a receiver in the ear canal. The hardest problem has always been noise. People with hearing loss usually struggle most in restaurants, parties and busy streets, and turning up the volume amplifies the clatter along with the voices. Traditional hearing aids tackled this with directional microphones, which favour sound from in front of the wearer, and rule-based noise reduction that turns down bands dominated by steady noise such as fans or engines. These help, but they struggle with babble, where the noise is other people talking and looks a lot like the speech you want. AI-based hearing aids add deep neural networks trained on large collections of recorded sound scenes. Oticon's More, launched in 2021, used a network trained on millions of sound scenes to balance speech and background. In 2024 Phonak's Audéo Sphere Infinio added a dedicated chip to run a denoising network that separates speech from noise in real time. Starkey has marketed on-device AI for sound classification and fall detection. Many devices also use classifiers that recognise the environment and change settings automatically. In the US, over-the-counter hearing aids became available after an FDA rule took effect in October 2022, for adults with perceived mild to moderate hearing loss. In 2024 the FDA also authorised a hearing aid feature for Apple's AirPods Pro 2. These can help many people, but they are not meant for severe or profound loss, children, sudden hearing loss, or ear pain and drainage, which need a clinician. A common misconception is that AI restores hearing. It cannot repair damaged hair cells in the inner ear. It processes sound more cleverly, and results still depend heavily on proper fitting.