聴覚知覚による音声評価の改善：CAPE-Vの教室およびオンライン併用トレーニング研究

Auditory-perceptual training is essential to reduce rating variability, yet most studies on hybrid training approaches are scarce, and most emphasize reliability rather than accuracy relative to expert standards.

This study evaluated a hybrid teaching tool combining in-class discussion with optional online practice (All-Voiced) for improving accuracy, reliability, and generalization of consensus auditory-perceptual evaluation of voice (CAPE-V) ratings in speech-language pathology students on both vowels and sentences.

Sixty-six graduate students completed a five-step CAPE-V training sequence, including baseline ratings, guided in-class discussion of prototypical samples, repeated ratings, optional online training, and a final test. Students rated Overall Severity, Roughness, and Breathiness relative to expert anchors. Accuracy (mean absolute error, % within 10 mm, and interquartile range [IQR]) and reliability (intraclass correlation coefficients) were analyzed for step, discussion, training, novelty, and sample type effects.

Learners improved significantly after in-class discussion (P < 0.001), with gains largely retained at final testing. Breathiness showed the largest improvement, approaching clinical minimally detectable change thresholds. Optional online training yielded modest additional benefits, significant only for Breathiness (P < 0.001) and for Overall Severity mean absolute error (P = 0.03). Performance generalized less effectively to novel than familiar voices (P ≤ 0.002), except for Overall Severity % within IQR (P = 0.47). Across Step, Discussion, Training, and Novelty conditions, sentences were generally rated more accurately than vowels: they showed consistently lower errors (mean absolute error, P ≤ 0.014) and higher accuracy within 10 mm (P ≤ 0.05), while % within IQR advantages emerged mainly for Roughness (Step and Discussion, P < 0.001) and Breathiness (Training and Novelty, P ≤ 0.002), but not for Overall Severity.

Hybrid training combining expert-led discussion with online practice improves both accuracy and reliability of CAPE-V ratings in novices. Even modest changes represent important pedagogical gains that support more consistent and valid clinical decision-making.