Barbera David S, Huckvale Mark, Fleming Victoria, Upton Emily, Coley-Fisher Henry, Doogan Catherine, Shaw Ian, Latham William, Leff Alexander P, Crinion Jenny
Institute of Cognitive Neuroscience, University College London, U.K.
Speech, Hearing & Phonetic Sciences, University College London, U.K.
Comput Speech Lang. 2021 Sep;69:None. doi: 10.1016/j.csl.2021.101221.
Anomia (word-finding difficulties) is the hallmark of aphasia, an acquired language disorder most commonly caused by stroke. Assessment of speech performance using picture naming tasks is a key method for both diagnosis and monitoring of responses to treatment interventions by people with aphasia (PWA). Currently, this assessment is conducted manually by speech and language therapists (SLT). Surprisingly, despite advancements in automatic speech recognition (ASR) and artificial intelligence with technologies like deep learning, research on developing automated systems for this task has been scarce. Here we present NUVA, an utterance verification system incorporating a deep learning element that classifies 'correct' versus' incorrect' naming attempts from aphasic stroke patients. When tested on eight native British-English speaking PWA the system's performance accuracy ranged between 83.6% to 93.6%, with a 10-fold cross-validation mean of 89.5%. This performance was not only significantly better than a baseline created for this study using one of the leading commercially available ASRs (Google speech-to-text service) but also comparable in some instances with two independent SLT ratings for the same dataset.
命名障碍（找词困难）是失语症的标志，失语症是一种后天性语言障碍，最常见的病因是中风。使用图片命名任务评估言语表现是诊断和监测失语症患者（PWA）对治疗干预反应的关键方法。目前，这项评估由言语和语言治疗师（SLT）手动进行。令人惊讶的是，尽管自动语音识别（ASR）和深度学习等人工智能技术取得了进展，但针对这项任务开发自动化系统的研究却很少。在此，我们展示了NUVA，这是一个话语验证系统，它包含一个深度学习元素，可对失语性中风患者的“正确”与“错误”命名尝试进行分类。在8名以英式英语为母语的PWA上进行测试时，该系统的性能准确率在83.6%至93.6%之间，10倍交叉验证均值为89.5%。这一性能不仅显著优于使用领先的商用ASR之一（谷歌语音转文本服务）为本研究创建的基线，而且在某些情况下与同一数据集的两个独立SLT评级相当。
