Objective: We examined the Spatial AI model running on the Fortell AI hearing aids to see whether it improves perceived ease of understanding in noisy, multitalker environments relative to hearing aids using more traditional processing.
Design: In a blinded, randomised-controlled trial, participants compared the investigational hearing aid to five control hearing aids, representing the latest technology from the five major hearing aid manufacturers. Recordings of the hearing aids in noisy, multitalker environments were presented to participants in a home environment via PC/headphones. In each of 100 randomised trials, participants rated which hearing aid made a target talker easier to understand. Preference data were compared to objective metrics.
Study sample: Twenty adults with mild to moderately-severe bilateral sensorineural hearing loss.
Results: The investigational device was preferred to the control hearing aids in 99 out of 100 trials, with one tie. The investigational hearing aid had an average SNR of 12.2 dB (over the unaided condition), compared to between 3.0 and 5.9 dB for the five control hearing aids.
Conclusions: For most clinicians, performance in noise is a top consideration when selecting a hearing device. The consistent preference for the investigational hearing aid suggests that Spatial AI may significantly improve perceived ease of understanding in challenging, multitalker noise.
Keywords: Speech-in-noise; deep neural network; hearing aids; listening effort; machine learning; speech enhancement.