Examining Acoustic Correspondence Between Artificial and Human Speech Patterns in Korean Language Instruction for Non-Native Speakers

This study examines the extent of acoustic similarity between artificial and human speech within the framework of Korean language education for non-native learners. The main objective is to ascertain whether the speech patterns of language learners resemble those of native speakers or those produced by artificial intelligence (AI) text-to-speech (TTS) systems. Utilizing a mixed-methods approach, fifteen experimental teams—consisting of native Korean speakers, Korean language learners from Universiti Malaysia Sabah, and five Korean TTS engines—generated a corpus of forty-five recorded utterances. Each sample was changed into spectral data and analyzed with Dynamic Time Warping (DTW) to find differences in timing and frequency.