Skip to main content
Article

Spoken Language Evaluation Through Aienhanced Pronunciation Feedback Systems

Nadira KhalkhmurzayevaHead of the Higher School of Japanese Studies, Tashkent State University of Oriental Studies,UzbekistanZarnigor SobirovaBukhara State University,Interfaculty Department of Foreign Languages,Bukhara,UzbekistanSevara MuhammadiyevaShahrisabz State Pedagogical Institute,Shahrisabz,UzbekistanNafisa NorovaShahrisabz State Pedagogical Institute,Shahrisabz,UzbekistanFeruza XasanovaShahrisabz State Pedagogical Institute,Shahrisabz,UzbekistanSohibjon MisirovNamangan state institute of foreign languages,Namangan,Uzbekistan,160123Shukurllo MamatovFaculty of Foreign Languages of the Jizzakh State Pedagogical University
2026
ABI

Abstract

Speech evaluation is one of the significant parts of language learning and assessment, which involves correct pronunciation and fluency analysis. As artificial intelligence has been improved, AI-enhanced pronunciation feedback systems have come into existence to give learners automated, real-time feedback. The current systems tend to use the traditional speech recognition algorithms, which can be unable to face the variability of accents, mispronunciations, and the difference in time of speech patterns and provide incorrect feedback and prevent effective learning of the learner. To cover these shortcomings, the offered approach incorporates Dynamic Time Warping (DTW) as the fundamental approach to the evaluation of pronunciation. DTW allows the accurate alignment of the speech of the learners to the reference utterances based on the temporal differences calculation and the reduction of the distance between the sequences of the speech features. By this method, any changes in speed of speech, intonation, and articulation can be handled effectively and precise feedback can be provided by the system. To evaluate pronunciation accuracy, identify instances of mispronunciation and provide language learners with the corrective recommendations, the proposed DTW-based system is implemented. The results of the experiment show that this method increases the accuracy of feedback, minimizes the errors of alignment, and improves pronunciation abilities of the learners compared to the conventional ones.

Topics

Identifiers

Citations and references

Cited by 013 references