VISUAL ARTICULATORY FEEDBACK

视觉关节反馈

• 批准号: 6213372
• 负责人: Joel M MacAuslan
• 金额: $ 13.53万
• 依托单位: SPEECH TECHNOLOGY/APPLIED RESEARCH CORP.
• 依托单位国家: 美国
• 财政年份: 2000
• 资助国家: 美国
• 起止时间: 2000-09-01 至 2001-03-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6213372
• 关键词: biomedical equipment development; communication disorder aid; computer program /software; computer system design /evaluation; deafness; speech; speech recognition; speechreading; visual feedback

A system providing visual feedback of articulation for people practicing English pronunciation would be useful to several populations. The proposed visual articulatory feedback system (VAF) will be based on articulatory x-ray microbeam data obtained in the last few years. VAF differs from current feedback systems based on automatic speech recognition (ASR) technology by its reliance on these data. Further, VAF does not require phonetic categorization to infer articulation, unlike ASR-based feedback. These factors will enable VAF to provide finer- grained articulatory feedback than current systems. VAF will also allow feedback for nonstandard pronunciations, such as those of dysarthric, deaf, and non-native speakers. The system will provide feedback with three planes of view, as well as a pseudopalatogram showing the teeth and palate. Given a student's speech sample, each view will show the corresponding articulator positions compared to the target. Thus, it will indicate the articulatory changes needed to produce better pronunciation. In Phase I, the research will focus on the difficult /ee/-/ih/ vowel distinction. In a multiple-subject case-study design, we will test the prototype against an existing auditory-feedback system. The two systems will be compared for efficiency and effectiveness in training deaf, dysarthric, and non-native speakers. PROPOSED COMMERCIAL APPLICATIONS: Potential applications include both self-paced and clinician-directed training courses for correction and maintenance of pronunciation. This applies particularly to adult populations who are deaf, are non-native speakers of English, or are diagnosed with a degenerative (dysarthria- producing) neuropathology such as Parkinsons disease or amyotrophic lateral sclerosis.

为练习英语发音的人提供清晰度的视觉反馈的系统将对若干人群有用。建议的视觉发音反馈系统（VAF）将基于发音的X射线微束数据在过去几年中获得的。VAF与当前基于自动语音识别（ASR）技术的反馈系统的不同之处在于它依赖于这些数据。此外，VAF不需要语音分类来推断发音，不像基于ASR的反馈。这些因素将使VAF能够提供比当前系统更细粒度的发音反馈。VAF还将允许对非标准发音的反馈，例如构音障碍者、聋人和非母语者的发音。该系统将提供三个视图平面的反馈，以及显示牙齿和腭的伪腭图。给定学生的语音样本，每个视图将显示与目标相比的相应发音架位置。因此，它将指示产生更好发音所需的发音变化。在第一阶段，研究将集中在困难的/ee/-/ih/元音的区别。在一个多主题的案例研究设计中，我们将测试原型对现有的决策反馈系统。这两个系统将进行比较的效率和有效性培训聋人，构音障碍，和非母语人士。拟议的商业应用：潜在的应用包括自定进度和临床医生指导的纠正和维持发音的培训课程。这尤其适用于耳聋、英语非母语者或被诊断患有退行性（产生构音障碍）神经病理学如帕金森病或肌萎缩侧索硬化症的成年人群。