SSPDT: A Multilevel, Model-Based Speech Perception Test

SSPDT：多级、基于模型的语音感知测试

• 批准号: 6338146
• 负责人: Joseph Bochner
• 金额: $ 10.08万
• 依托单位: P.A.I.S., INC.
• 依托单位国家: 美国
• 财政年份: 2001
• 资助国家: 美国
• 起止时间: 2001-05-11 至 2002-04-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6338146
• 关键词: auditory discrimination; hearing disorders; hearing tests; human subject; mathematical model; psychometrics; sound perception; speech recognition; technology /technique development

DESCRIPTION: The broad goal of the proposed research is development and commercialization of a multilevel model-based speech discrimination testing system designed for individuals experiencing speech discrimination difficulties ranging from mild to severe. Development of the testing system benefits from related research conducted over a 10-year period which resulted in a prototypic speech discrimination test for persons with severe to profound hearing losses. In the proposed work, we seek to increase the difficulty of the discrimination tasks and decrease test administration time. We perform detailed item, reliability and validity analyses to establish the empirical foundation fro a computerized, adaptive testing system responsive to the individual differences which exist among persons receiving audiological services. The test materials consist of sentence-level utterances presented to subjects with and without background noise. Fifty subjects will participate in the study. The testing system will be evaluated in terms of established psychometric criteria, and also for goodness-of-fit to a mathematical model for person measurement. This research represents and interdisciplinary collaboration. It is technically sophisticated in terms of current audiological practice, and has significant advantages over conventional methods of speech discrimination testing.

描述：拟议研究的总体目标是开发和 基于多级模型的语音辨别测试的商业化 专为有言语辨别困难的个人设计的系统 从轻微到严重。测试系统的开发受益于 相关研究历时 10 年，最终形成了原型 针对重度至极重度听力损失者的言语辨别测试。 在拟议的工作中，我们寻求增加歧视的难度 任务并减少测试管理时间。进行详细的项目， 信度和效度分析为建立实证基础 适应个体差异的计算机化自适应测试系统 存在于接受听力服务的人中。测试材料 由呈现给受试者的句子级话语组成，有和没有 背景噪音。五十名受试者将参与这项研究。测试 将根据既定的心理测量标准对系统进行评估，并且 还用于人员测量数学模型的拟合优度。这 研究代表跨学科合作。从技术上来说是 就当前的听力学实践而言，非常复杂，并且具有重要意义 与传统的言语辨别测试方法相比的优点。