Acoustic analysis in stuttering: a machine-learning study

Abstract

BackgroundStuttering is a childhood-onset neurodevelopmental disorder affecting speech fluency. The diagnosis and clinical management of stuttering is currently based on perceptual examination and clinical scales. Standardized techniques for acoustic analysis have prompted promising results for the objective assessment of dysfluency in people with stuttering (PWS).ObjectiveWe assessed objectively and automatically voice in stuttering, through artificial intelligence (i.e., the support vector machine – SVM classifier). We also investigated the age-related changes affecting voice in stutterers, and verified the relevance of specific speech tasks for the objective and automatic assessment of stuttering.MethodsFifty-three PWS (20 children, 33 younger adults) and 71 age−/gender-matched controls (31 children, 40 younger adults) were recruited. Clinical data were assessed through clinical scales. The voluntary and sustained emission of a vowel and two sentences were recorded through smartphones. Audio samples were analyzed using a dedicated machine-learning algorithm, the SVM to compare PWS and controls, both children and younger adults. The receiver operating characteristic (ROC) curves were calculated for a description of the accuracy, for all comparisons. The likelihood ratio (LR), was calculated for each PWS during all speech tasks, for clinical-instrumental correlations, by using an artificial neural network (ANN).ResultsAcoustic analysis based on machine-learning algorithm objectively and automatically discriminated between the overall cohort of PWS and controls with high accuracy (88%). Also, physiologic ageing crucially influenced stuttering as demonstrated by the high accuracy (92%) of machine-learning analysis when classifying children and younger adults PWS. The diagnostic accuracies achieved by machine-learning analysis were comparable for each speech task. The significant clinical-instrumental correlations between LRs and clinical scales supported the biological plausibility of our findings.ConclusionAcoustic analysis based on artificial intelligence (SVM) represents a reliable tool for the objective and automatic recognition of stuttering and its relationship with physiologic ageing. The accuracy of the automatic classification is high and independent of the speech task. Machine-learning analysis would help clinicians in the objective diagnosis and clinical management of stuttering. The digital collection of audio samples here achieved through smartphones would promote the future application of the technique in a telemedicine context (home environment).

Computational Intelligence-Based Stuttering Detection: A Systematic Review

“…In the literature, researchers have applied various datasets, including University College London's Archive of Stuttered Speech (UCLASS) [11], SEP-28k [12], FluencyBank [13], and LibriStutter [14], and have also used resources like VoxCeleb [15] (see Table 1). Furthermore, some researchers such as [16][17][18] have even made their own customized datasets to cater to their specific research needs. This combined endeavor emphasizes the importance of carefully curated and extensive data in advancing the field of stuttering-detection technology via deep learning and AI methods.…”

“…Furthermore, several studies [16][17][18] chose to create their own tailored datasets to address their specific research objectives. For instant, in [16], the dataset comprised 20 individuals aged between 15 and 35, all of whom had been diagnosed with stuttering by qualified speech language pathologists.…”

“…Asci et al [18] recruited 53 individuals with stuttering (24 females and 29 males aged 7-30) alongside 71 age-and sex-matched controls (29 females and 44 males aged 7-30). All participants were native Italian speakers, non-smokers, and had no cognitive or mood impairments, hearing loss, respiratory disorders, or other conditions that could affect their vocal cords.…”

“…Some studies [14,[16][17][18][19][20][21][22]26,27] used just one dataset, whereas others [6,[23][24][25] opted for multiple datasets. In one particular study [6], the researchers went a step further by enhancing their data with the MUSAN dataset for added diversity.…”

“…Contrarily, part-word repetition and syllable repetition were the least discussed, having only a single mention each. The predominant stuttering categories, namely repetition, prolongation, and block, were discussed in previous works [6,16,18,22,24]. Other variations of stuttering, which can be considered subcategories of the primary classes, were addressed in previous studies [14,19,20], including sound repetitions, word repetitions, and phrase repetitions.…”