iHearBetter-in-Motion: Unlocking communication in motion for hearing-impaired individuals

听力损失是英国第二大常见残疾，影响1100万人（1/6;听力损失行动，2015）。随着人口老龄化，到2035年，估计英国将有超过1560万人患有听力损失（1/5; Action on Hearing Loss，2015）。助听器将音频信号（例如，助听器的麦克风）转换为磁信号，然后由具有听力环路设置的助听器检测并选择性地放大。因此，向听力受损个体（HII）呈现感兴趣的声音（例如，在音乐会期间），没有音频干扰（反馈或背景噪声）。然而，听力环技术无法应用于自然对话，HIIs只能使用放大周围所有声音的设备，因此不适合在嘈杂环境中多人同时说话的环境中使用。今天，虽然英国有670万人可以从助听器中受益，但只有200万人经常使用助听器（Hearing Link，2019）。多项研究发现，人们选择不戴助听器的主要原因之一是他们在嘈杂环境中的表现不佳（McCormack和Fortnum，2013）;“鸡尾酒会效应”。因此，75岁以上的HIIs中有35%以上假装他们能听到东西，25%感到孤立（皇家志愿服务，2018）。这不仅仅是老龄化社区的问题;超过450万的HIIs处于工作年龄，其中75%的人认为他们的就业机会受到听力损失的限制（听力损失行动，2015）。听力损失对英国经济造成的经济负担估计为每年250亿英镑的生产力损失和失业（布鲁内尔大学伦敦，2015年;阿赫博尔德等人，2014).在Innovate UK的资助下（编号46691），AudioTelligence和剑桥大学开发了改变游戏规则的辅助听力技术，该技术可以将语音/感兴趣的声音与真实的背景噪音分开。我们的方法融合了AudioTelligence的专利音频源分离技术（WO 2019016494）与剑桥大学的尖端听觉注意力研究。通过将音频源分离算法与听觉注意力线索相结合，我们可以自动选择并向最终用户呈现所需的音频源，从而减少/消除背景噪声。然而，我们现有的方法仅适用于静态情况;例如，当HIIs坐在会议或餐厅中时。在这里，我们将开发新的算法来扩展我们的方法，以涵盖HII正在移动的情况，为HII在工作和社交时解锁通信。

Hearing loss is the second most common disability in the UK, affecting 11M people (1 in 6; Action on Hearing Loss, 2015). With an ageing population, by 2035 it is estimated more than 15.6M people in the UK will have hearing loss (1 in 5; Action on Hearing Loss, 2015).Hearing loops convert sound from electrical sources (e.g., loop system's microphone) into a magnetic signal, which is detected and selectively amplified by hearing aids with hearing-loop settings. Thus, hearing-impaired individuals (HIIs) are presented with the sound-of-interest (e.g., during a concert), without audio interference (feedback or background noise). However, hearing-loop technology cannot be applied in natural conversations and HIIs are left with devices that amplify all surrounding sounds, making them unsuitable for use in environments where more than one person is speaking at the same time in a noisy environment.Today, while 6.7M people in the UK could benefit from hearing aids, only 2M people use them regularly (Hearing Link, 2019). Multiple studies have found that one of the main reasons for people choosing not to wear hearing aids is their poor performance in noisy situations (McCormack and Fortnum, 2013); "cocktail-party effect".Consequently, 35%+ of HIIs aged 75+ get through social occasions pretending they can hear things and 25% feel isolated (Royal Voluntary Service, 2018). This is not just an issue in aging communities; over 4.5M HIIs are working age and c75% of these feel their employment opportunities have been limited by their hearing loss (Action on Hearing Loss, 2015). The economic burden of hearing loss to the UK economy has been estimated at £25 billion/year in lost productivity and unemployment (Brunel University London, 2015; Archbold et al., 2014).With Innovate UK funding (no. 46691), AudioTelligence and the University of Cambridge have developed game-changing assistive listening technology that can separate the voice/sound-of-interest from background noise in real time. Our approach fuses AudioTelligence's patented audio source-separation technologies (WO2019016494) with the University of Cambridge's cutting-edge auditory attention research. By combining audio source-separation algorithms with auditory attention cues, we can automatically select and present the desired audio source to the end user, reducing/eliminating background noise.However, our existing approach is only applicable to static situations; for example, when HIIs are sat down in meetings or in a restaurant. Here, we will develop new algorithms to extend our approach to encompass situations where HIIs are moving, unlocking communication for HIIs at work and when socialising.