Speech-controlled automated matrix test (Knowledge Transfer Project) (T01#)

语音控制自动矩阵测试（知识转移项目）（T01

• 批准号: 396657276
• 金额: --
• 依托单位: Abteilung Medizinische Physik
• 依托单位国家: 德国
• 项目类别: CRC/Transregios (Transfer Project)
• 财政年份: 2018
• 资助国家: 德国
• 起止时间: 2017-12-31 至 2022-12-31
• 项目状态: 已结题
• 来源: https://gepris.dfg.de/gepris/projekt/396657276?language=en
• 关键词: Speech; controlled; automated; matrix; test

In a society in which more than 17% of the population suffer from a hearing deficit, reliable and easily accessible hearing test methods become more and more important to allow for an earlyintervention (e.g. with listening devices) and to assess the benefit of such interventions. Matrix sentence tests are an exact and internationally accepted method, but up to now require the presence of a human supervisor who logs the listener’s responses. We propose to develop aprototype for a Speech-controlled Automated Matrix Test (SAMT) as an easily accessible hearingtest, which is based on automatic speech recognition (ASR). The system will enable unsupervisedhearing testing which will significantly raise its applicability since it requires far less resources than its supervised counterpart. SAMT will be based on the Oldenburg Sentence test (OLSA), which is available in 16 languages. Through this project, we will provide a system that can perform the test in two languages, as well as a development kit for rapid adaptation to new languages. In a clinical setting, the system will enable a reliable clinical hearing test even when the supervisor does not speak the listener’s language, thereby lifting another restriction of today’s listening tests. The project builds on a successful 12-year funding of fundamental research on robust automatic speech recognition as a model for auditory perception of normal and hearing-impaired listeners. Hence, the project will transfer the knowledge and techniques into applications for daily life with the aim to alleviate the situation for persons with a hearing loss.

在一个超过17%的人口患有听力缺陷的社会中，可靠且易于获得的听力测试方法变得越来越重要，以允许早期干预（例如使用听力设备）并评估此类干预的益处。矩阵句子测试是一种精确的、国际公认的方法，但到目前为止，它需要一个记录听众反应的人类监督者在场。本文提出了一种基于自动语音识别（ASR）的语音控制自动矩阵测试（SAMT）的原型系统。该系统将实现无监督的听力测试，这将大大提高其适用性，因为它需要的资源远远少于有监督的测试。SAMT将基于Oldenburg句子测试（OLSA），该测试有16种语言版本。通过这个项目，我们将提供一个可以用两种语言进行测试的系统，以及一个快速适应新语言的开发工具包。在临床环境中，该系统将实现可靠的临床听力测试，即使主管不会说听者的语言，从而解除了当今听力测试的另一个限制。该项目建立在一个成功的12年资助的基础研究强大的自动语音识别作为一个模型的听觉感知的正常和听力受损的听众。因此，该项目将把知识和技术转移到日常生活中的应用中，以减轻听力损失者的处境。