Рақамли технологияларнинг назарий ва амалий масалалари Ҷилди 9 № 1 (2026) · Саҳифаҳои 64-79
Neural network based linguistic trainer for the diagnosis and correction of speech disorders
Aбдуллaевa, М.И., Каримова, М.Б.
Аннотатсия
Today, the development of speech correction systems for general speech underdevelopment in preschool-age children using speech technologies is a pressing and pressing priority. This article describes the structure of a linguistic trainer, including diagnostic modules based on fuzzy logic, interpretation and understanding of user requests, and response generation based on NLU approaches, speech synthesis from text to build a convenient relationship between the user and the system, adaptive and specialized exercises using pedagogical technologies, and an analytical block based on specialized metrics such as progress metrics, rate of progress, speech unit acquisition metrics, and attendance metrics, which collectively represent the conclusions of a speech therapist's AI. Each module is described in terms of both its structure and effective approaches. Literary sources are cited for detailed results of neural network models of the linguistic trainer. Based on the analysis, it was found that the rate of progress metric calculated by the linguistic trainer is 1.2%, while traditional rehabilitation is 0.6%. When using the simulator, the rate of skill formation within the speech unit under consideration increased on average by ~1.7 times, and the indicator of absolute increase in pronunciation accuracy (ΔP) of mastering the speech unit, which indicates that over the same period of speech rehabilitation when using the simulator, an acceleration of the process of 4.5–5 times is achieved.
оценкa произношениямодуль генерaции речиNLU-модульцифровaя логопедиялингвистический тренaжёрpronunciation assessmentspeech formation moduleNLU moduledigital speech therapylinguistic trainer
Манбаи метамаълумот: бойгонии OAI-PMH-и маҷалла · Sindex матни пурраро нигоҳ намедорад, ба манбаъ пайванд медиҳад.