연속 문장 발화에서 사전 정의된 청크 검출에 대한 인공지능 기반의 후향적 분석 연구
Deep-Learning-Based Segmentation of Predefined Chunks in Connected Speech: A Retrospective Analysis
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Background and Objectives In institutional settings, manually segmenting connected speech is a time-consuming and labor-intensive process. This study aims to develop a deep-learning model for automating this process, evaluating its accuracy, and determining the minimum dataset size for effective performance. Materials and Method Voice data from 524 individuals with pathological conditions and 502 individuals with normal conditions, totaling 1026 samples, were used. Each voice sample had 17 chunks, including a “summer” sentence (15 chunks) and vowels /α/ and /i/. The deep-learning model employed in this study is based on the multi-layer perceptron-mixer architecture. This study evaluated performance using the Intersection over Union (IoU) metric, commonly employed in artificial intelligence-based image detection for chunk segmentation. Results The accuracy of chunk identification at the frame level was 96.47%. Using IoU metrics, chunk segmentation accuracy was 98.15% at IoU ≥0.6, 96.03% at IoU ≥0.7, and 89.78% at IoU ≥0.8. Optimal dataset size exploration indicated that more than 700 connected speech datasets were needed for successful training, maintaining F1-scores up to 95% at IoU ≥0.7. Conclusion The artificial intelligence model is suitable for the development of an automated system that efficiently divides segments in the institutional collection of voice data. This suggests its potential utility in advancing voice research using connected speech
Abstract
Background and Objectives In institutional settings, manually segmenting connected speech is a time-consuming and labor-intensive process. This study aims to develop a deep-learning model for automating this process, evaluating its accuracy, and determining the minimum dataset size for effective performance. Materials and Method Voice data from 524 individuals with pathological conditions and 502 individuals with normal conditions, totaling 1026 samples, were used. Each voice sample had 17 chunks, including a “summer” sentence (15 chunks) and vowels /α/ and /i/. The deep-learning model employed in this study is based on the multi-layer perceptron-mixer architecture. This study evaluated performance using the Intersection over Union (IoU) metric, commonly employed in artificial intelligence-based image detection for chunk segmentation. Results The accuracy of chunk identification at the frame level was 96.47%. Using IoU metrics, chunk segmentation accuracy was 98.15% at IoU ≥0.6, 96.03% at IoU ≥0.7, and 89.78% at IoU ≥0.8. Optimal dataset size exploration indicated that more than 700 connected speech datasets were needed for successful training, maintaining F1-scores up to 95% at IoU ≥0.7. Conclusion The artificial intelligence model is suitable for the development of an automated system that efficiently divides segments in the institutional collection of voice data. This suggests its potential utility in advancing voice research using connected speech
참고문헌 (19)
- Sell D, "Issues in perceptual speech analysis in cleft palate and related disorders : a review", Int J Lang Commun Disord, 40(2), pp.103~121 (2005)
- Boschi V, "Connected speech in neurodegenerative language disorders : a review", Front Psychol, 8, pp.269~ (2017)
- Corcoran CM, "Using language processing and speech analysis for the identification of psychosis and other disorders", Biol Psychiatry Cogn Neurosci Neuroimaging, 5(8), pp.770~779 (2020)
- Seok J, "Artificial intelligence for clinical research in voice disease", J Korean Soc Laryngol Phoniatr Logop, 33(3), pp.142~155 (2022)
- Hintze JM, "Spasmodic dysphonia : a review. Part 2 : characterization of pathophysiology", Otolaryngol Head Neck Surg, 157(4), pp.558~564 (2017)
- Syed SA, "Meta-analysis of voice disorders databases and applied machine learning techniques", Math Biosci Eng, 17(6), pp.7958~7979 (2020)
- Mesallam TA, "Development of the Arabic voice pathology database and its evaluation by using speech features and machine learning algorithms", J Healthc Eng, 2017, pp.8783751~ (2017)
- Kay Elemetrics Corporation, "Disordered voice database, model 4337(version 1.03)", Kay Elemetrics Corporation, ~ (1994)
- Woldert-Jokisz B, "Saarbruecken voice database" (2007), Institute of Phonetics, Saarland University
- Zabidi A, "Detection of asphyxia in infants using deep learning convolutional neural network(CNN)trained on Mel frequency cepstrum coefficient(MFCC)features extracted from cry sounds", J Fundam Appl Sci, 9(3S), pp.768~778 (2017)
- Kim H, "Convolutional neural network classifies pathological voice change in laryngeal cancer with high accuracy", J Clin Med, 9(11), pp.3415~ (2020)
- Hu HC, "Deep learning application for vocal fold disease prediction through voice recognition : preliminary development study", J Med Internet Res, 23(6), pp.e25247~ (2021)
- Xie X, "A voice disease detection method based on MFCCs and shallow CNN", J Voice (2023)[DOI]
- Zhang J, "Automatic detection of obstructive sleep apnea events using a deep CNN-LSTM model", Comput Intell Neurosci, 2021, pp.5594733~ (2021)
- Yu H, "A sleep apneahypopnea syndrome automatic detection and subtype classification method based on LSTM-CNN", Biomed Signal Process Control, 71, pp.103240~ (2022)
- Ullah I, "An automated system for epilepsy detection using EEG brain signals based on deep learning approach", Expert Syst Appl, 107, pp.61~71 (2018)