Peer-reviewed article
Application of machine learning techniques to automate speech mispronunciation detection in children and adolescents.
Ameli N, Ravesh MN, Dalmazo LMT, Pollock K, DeSantis D et al. ·
Summary
Written by AI, not yet AICS's view · claude-haiku-4-5-20251001This is a technical study on developing a machine learning tool to automatically detect speech sound disorders in children aged 6-18 years. The researchers trained a deep learning model on 1,800 standardized word recordings and achieved 90.9% accuracy in identifying mispronunciation errors. The tool shows promise as a screening aid where specialist access is limited, but the study is narrowly focused on algorithm validation rather than evaluating real-world child health or well-being outcomes.
Key points
- Speech sound disorders affected 16.15% of recordings in the sample, with /θ/ and /ð/ sounds most commonly mispronounced
- The machine learning model achieved 90.9% accuracy, 77.8% sensitivity, and 98.2% specificity on held-out test data
- Strong inter-rater agreement (Cohen's κ = 0.84) confirmed reliable expert labeling of mispronunciation
- No significant differences in speech errors were found by sex, age group, or malocclusion status
- Authors note the tool could support triage in settings with limited specialist access but needs validation in broader clinical settings
Limitations
This is a technical validation study on a single standardized elicitation protocol in a controlled laboratory setting. It does not evaluate whether the tool improves actual clinical outcomes, reduces barriers to care, or how it performs in diverse real-world contexts with background noise, different languages, or non-standardized speech samples.
Abstract
Introduction Speech sound disorders (SSDs) are common in children and can affect academic and psychosocial outcomes. Clinical identification relies on expert auditory-perceptual assessment, which is time-intensive and may vary across raters. Automated screening tools could support triage when access to specialists is limited. Objective This study aimed to develop and evaluate a deep learning (DL) classifier for detecting mispronunciation errors in standardized pediatric speech recordings obtained using a structured fricative-focused word elicitation protocol, using expert-adjudicated labels as the reference standard. Methodology In this cross-sectional study, we analyzed 100 participants (6-18 years) providing 1,800 standardized word recordings. Two expert speech-language pathologists (SLPs) labeled recordings as mispronunciation present vs. absent; disagreements were adjudicated by a third SLP. Inter- and intra-rater reliability were assessed. Audio was denoised and standardized to 16 kHz. A pretrained transformer speech model (WavLM Base+) was fine-tuned for recording-level binary classification. Class imbalance was addressed using augmentation, weighted sampling, and focal loss. Performance was assessed on a held-out test set at the recording level using accuracy, sensitivity, specificity, precision, F1-score, and area under the ROC curve (AUC). Results The overall recording-level prevalence of speech sound errors was 16.15%, with the "th" category (/θ/ + /ð/) being the most frequently mispronounced phoneme group. No significant differences were observed by sex, age group (6-9, 10-13, 14-18 years), or malocclusion status (p > 0.05). Labeling showed strong agreement (Cohen's κ = 0.84; intra-rater reliability 0.89-0.92). On the test set (253 recordings), the model achieved 90.9% accuracy, 77.8% sensitivity, 98.2% specificity, 95.9% precision, and AUC = 0.936. Conclusions Fine-tuned pretrained speech representations demonstrated promising performance for screening pediatric mispronunciation errors under standardized recording conditions using expert labels. The high-specificity profile supports its use as a screening and triage decision-support tool, while future studies are needed to validate performance across broader clinical and real-world settings.