聲帶內收肌肌張力障礙中語言學導向的自動化喀擦聲估計

// Linguistically Informed Automated Estimates of Creak in Adductor Laryngeal Dystonia

本研究探討非常規語境下的「Creаky Voice」（喉音沙啞）能否更有效區分「聲帶肌痙攣性失調症」（AdLD）與正常聲音｜檢測中，50名患者與50名對照者朗讀文章，利用演算法評估預期與非預期喉音沙啞｜結果顯示非預期喉音沙啞對AdLD辨識更具特異性，診斷準確率達0.82（AUC），強調語境考量有助提升診斷標記的準確性。

// This study examined whether creaky voice outside typical linguistic contexts better differentiates adductor laryngeal dystonia (AdLD) from normal voices. Fifty speakers with AdLD and 50 controls read a passage while an algorithm measured expected and unexpected creak. Unexpected creak was more distinctive of AdLD and showed higher diagnostic accuracy (AUC = .82) than total creak. The findings suggest considering linguistic context improves acoustic markers for diagnosing AdLD.

Published 2026年8月25日