Automatic Recognition of Prolongations and Repetitions in Stuttering Speech using ANN

摘要

Stuttering is a disturbance in normal fluent flow of speech unintentionally by disfluencies such as sound/syllable repetition, prolongation, or blocks. Stuttering is a fluency disorder affecting millions of people in their day to day life. The flow of speech is disrupted by stuttering. Automatic recognition of speech disfluencies in persons with stuttering is a great aid to admit the challenges and is a prominent technology involving human computer interaction. There are high proportions of prolongations and repetitions in stuttering compared to the other types of disfluencies such as pauses, interjections, and hesitations. This work is intended for automatic recognition of instances of prolongations and repetitions in stuttering. Automatic detection of prolongations and repetitions greatly facilitate the assessment of stuttering. The UCLASS database is utilized for the analysis of repetitions and prolongations in the speech of persons with stuttering (PWS). Instances of prolongations and repetitions in stuttering are recognized by using Artificial Neural Networks (ANN). The experimental investigations reveal that the proposed method shows promising results in identifying the instances of prolongations and repetitions in stuttering.