An Artificial Intelligence-Based Algorithm for the Assessment of Substitution Voicing

The purpose of this research was to develop an artificial intelligence-based method for evaluating substitution voicing (SV) and speech following laryngeal oncosurgery. Convolutional neural networks were used analyze spoken audio sources. A Mel-frequency spectrogram employed as input the deep network architecture. program trained using a collection 309 digitized recordings. acoustic index (ASVI) model elaborated regression analysis. This then tested with samples that unknown algorithm, results compared auditory-perceptual SV evaluation provided by medical professionals. statistically significant, strong correlation rs = 0.863 (p 0.001) observed between ASVI performed laryngologists. one-way ANOVA showed significant differences in control, cordectomy, partial laryngectomy, total laryngectomy patient groups < 0.001). lightweight algorithm reached rapid response rates 3.56 ms. provides fast efficient option patients after are comparable