Restoring speech intelligibility for hearing aid users with deep learning

Abstract

Almost half a billion people world-wide suffer from disabling hearing loss. While hearing aids can partially compensate for this, a large proportion of users struggle to understand speech in situations with background noise. Here, we present a deep learning-based algorithm that selectively suppresses noise while maintaining speech signals. The algorithm restores speech intelligibility for hearing aid users to the level of control subjects with normal hearing. It consists of a deep network that is trained on a large custom database of noisy speech signals and is further optimized by a neural architecture search, using a novel deep learning-based metric for speech intelligibility. The network achieves state-of-the-art denoising on a range of human-graded assessments, generalizes across different noise categories and—in contrast to classic beamforming approaches—operates on a single microphone. The system runs in real time on a laptop, suggesting that large-scale deployment on hearing aid chips could be achieved within a few years. Deep learning-based denoising therefore holds the potential to improve the quality of life of millions of hearing impaired people soon.

Hearing aid with artificial intelligence

“…In Starkey's Hearing Reality Sound AEC system, there are eight automatic sound classes: music, speech in quiet, speech in loud noise, speech in noise, machine, wind, noise, and quiet. (7) It promotes speech understanding in noisy environments by making accurate adjustments in gain, compression, directionality, noise management, and other parameters appropriate for each distinct class.…”

Socio-Technical Trust For Multi-Modal Hearing Assistive Technology

“…The design of hearing assistive devices involves the application and combination of a range of technologies from different domains, including speech enhancement, speaker identification, speech recognition, sound localization, and noise cancellation, among others [4,11]. These capabilities rely on a variety of inputs, such as audio, visual, and haptic (touch) feedback, where each could potentially be developed by different organizations.…”

“…In order to improve the power and understanding of AI algorithms that are integrated with hearing aids, deep learning based models are being used to amplify the performance and experience of the technology [4,10,23]. However, these deep models are black boxes and do not offer explanations for the decisions made by the device.…”

“…According to the World Health Organization (WHO), as of 2023 more than 5% of the world's population is deaf or hard of hearing (DHH) and this is expected to increase to 1 in 10 people worldwide by 2050 1 . Likewise, it is important to note that wearables [1], and more specifically earables [2] are entering the industrial tech market at a rapid rate, in an effort to offer more services to the general public -these range from background noise cancellation [3] and speech enhancement [4], to higher quality phone calls and hands-free music listening experiences. Wearables are essentially wearable computers that can perform tasks or provide services in a discrete manner that does not interrupt daily life [5].…”

The Exploratory Study on the Development of Online Based Auditory Training Contents: Based on the Analysis of the Application for Auditory-Training for the Elderly in their 70s and Older

“…특히, 많은 난청 노인들은 보청기를 착용한 후 에도 배경소음이 있는 곳에서 의사소통의 어려움이 있다 (Ding et al, 2023). 이러한 어려움의 해결을 위하여 배경소음을 줄 여주고 어음을 강화하는 등의 보청기 탑재 기술이 개발되었고 (Boymans & Dreschler, 2000;Picou et al, 2014), 최근에는 딥러닝 기술로 배경소음을 줄여주는 기술 (Healy et al, 2021;Zhao et al, 2018)까지 보청기 내에 탑재되었지만 여전히 소음 하에서는 의사소통의 어려움이 있다 (Diehl et al, 2023). 이는 난청과 노화로 인한 인지력의 저하로 인한 결과로 인해 발생한다 는 연구 결과 (Ghisletta et al, 2023)가 있으며 이를 개선하는 방법으로 청능훈련이 있다 (Lazard et al, 2023).…”