http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
손계원(Gyewon Son),소준섭(Junseop So),고주은(Jooeun Ko),이진우(Jin-Woo Lee),이정록(JeongRok Lee),신원선(Won-Sun Shin) 한국디지털콘텐츠학회 2024 한국디지털콘텐츠학회논문지 Vol.25 No.2
Child speech recognition has emerged as a significant research topic in the fields of human-computer interaction and educational technology. Childrens utterances possess distinct characteristics from adults, often making it challenging for conventional automatic speech recognition (ASR) models to accurately recognize their speech. In this study, we utilized OpenAIs Whisper model to transcribe the voices of 4-7 year-old children into text. Specifically, considering the differences in speech between children and adults, we conducted data refinement and dataset construction to enhance the models performance. These efforts present an approach to enhance the performance of the Whisper model for child speech recognition from the perspective of training data. Our method improved the error rate of Korean child voice recognition by 84%.