Tzu-Pei Tsai

Mr. Tzu-Pei Tsai

Indiana University Bloomington, Bloomington, California, USA
orcid

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 June 2021 in Journal of Medical Internet Research
Verified
Journal Article
Open access
HHHao-Chun Hu
Hao-Chun Hu
SCShyue-Yih Chang
Shyue-Yih Chang
CWChuen-Heng Wang
Chuen-Heng Wang
KLKai-Jun Li
Kai-Jun Li
HCHsiao-Yun Cho
Hsiao-Yun Cho
Oscar Kuang-Sheng LeeOscar Kuang-Sheng Lee
Oscar Kuang-Sheng Lee
Background: Dysphonia influences the quality of life by interfering with communication. However, a laryngoscopic examination is expensive and not readily accessible in primary care units. Experienced laryngologists are required to achieve an accurate diagnosis.Objective: This study sought to detect various vocal fold diseases through pathological voice recognition using artificial intelligence.Methods: We collected 189 normal voice samples and 552 samples of individuals with voice disorders, including vocal atrophy (n=224), unilateral vocal paralysis (n=50), organic vocal fold lesions (n=248), and adductor spasmodic dysphonia (n=30). The 741 samples were divided into 2 sets: 593 samples as the training set and 148 samples as the testing set. A convolutional neural network approach was applied to train the model, and findings were compared with those of human specialists.Results: The convolutional neural network model achieved a sensitivity of 0.66, a specificity of 0.91, and an overall accuracy of 66.9% for distinguishing normal voice, vocal atrophy, unilateral vocal paralysis, organic vocal fold lesions, and adductor spasmodic dysphonia. Compared with the accuracy of human specialists, the overall accuracy rates were 60.1% and 56.1% for the 2 laryngologists and 51.4% and 43.2% for the 2 general ear, nose, and throat doctors.Conclusions: Voice alone could be used for common vocal fold disease recognition through a deep learning approach after training with our Mandarin pathological voice database. This approach involving artificial intelligence could be clinically useful for screening general vocal fold disease using the voice. The approach includes a quick survey and a general health examination. It can be applied during telemedicine in areas with primary care units lacking laryngoscopic abilities. It could support physicians when prescreening cases by allowing for invasive examinations to be performed only for cases involving problems with automatic recognition or listening and for professional analyses of other clinical examination results that reveal doubts about the presence of pathologies.

17 August 2022 in Journal of Clinical Medicine
Verified
Journal Article
Open access
PDF file
ACAndy Wei-Ge Chen
Andy Wei-Ge Chen
CCChih-Hua Chen
Chih-Hua Chen
TLTSAI-MING LIN
TSAI-MING LIN
ACAngela Chih-Hui Chang
Angela Chih-Hui Chang
TTTzu-Pei Tsai
Tzu-Pei Tsai
SCShyue-Yih Chang
Shyue-Yih Chang
Unilateral vocal fold paralysis (UVFP) is a common cause of incomplete glottic closure, leading to significant somatic and social disabilities. Office-based autologous fat injection laryngoplasty (AFIL) has been proposed as an effective treatment for glottic insufficiency but has not been well-studied for UVFP. We enrolled 23 patients who underwent office-based structural AFIL due to unilateral vocal paralysis at our institution between February 2021 and January 2022. In the procedure, autologous fat was harvested and injected into the vocal fold under the guidance of flexible digital endoscopy for structural fat grafting. The voice handicap index-10 (VHI-10) score and perceptual voice measurements were collected before the operation, 2 weeks postoperatively, and 3 months postoperatively. Twenty-two patients were followed-up for at least 3 months. The VHI-10 score improved significantly from 29.65 ± 8.52 preoperatively to 11.74 ± 7.42 at 2 weeks (p < 0.0001) and 5.36 ± 6.67 at 3 months (p < 0.0001). Significant improvements in grades of dysphonia (p < 0.0001), breathiness (p < 0.0001), and asthenia (p = 0.004) were also noted at 3 months postoperatively when perceptual measurements were investigated. Office-based structural AFIL is an effective treatment for improving voice-related disability for UVFP patients.

25 October 2020 in JMIR Preprints
Verified
Preprint
Open access
HHHao-Chun Hu
Hao-Chun Hu
SCShyue-Yih Chang
Shyue-Yih Chang
CWChuen-Heng Wang
Chuen-Heng Wang
KLKai-Jun Li
Kai-Jun Li
HCHsiao-Yun Cho
Hsiao-Yun Cho
Oscar Kuang-Sheng LeeOscar Kuang-Sheng Lee
Oscar Kuang-Sheng Lee
BACKGROUND Dysphonia influences the quality of life by interfering with communication. However, a laryngoscopic examination is expensive and not readily accessible in primary care units. Experienced laryngologists are required to achieve an accurate diagnosis. OBJECTIVE This study sought to detect various vocal fold diseases through pathological voice recognition using artificial intelligence. METHODS We collected 189 normal voice samples and 552 samples of individuals with voice disorders, including vocal atrophy (n=224), unilateral vocal paralysis (n=50), organic vocal fold lesions (n=248), and adductor spasmodic dysphonia (n=30). The 741 samples were divided into 2 sets: 593 samples as the training set and 148 samples as the testing set. A convolutional neural network approach was applied to train the model, and findings were compared with those of human specialists. RESULTS The convolutional neural network model achieved a sensitivity of 0.66, a specificity of 0.91, and an overall accuracy of 66.9% for distinguishing normal voice, vocal atrophy, unilateral vocal paralysis, organic vocal fold lesions, and adductor spasmodic dysphonia. Compared with the accuracy of human specialists, the overall accuracy rates were 60.1% and 56.1% for the 2 laryngologists and 51.4% and 43.2% for the 2 general ear, nose, and throat doctors. CONCLUSIONS Voice alone could be used for common vocal fold disease recognition through a deep learning approach after training with our Mandarin pathological voice database. This approach involving artificial intelligence could be clinically useful for screening general vocal fold disease using the voice. The approach includes a quick survey and a general health examination. It can be applied during telemedicine in areas with primary care units lacking laryngoscopic abilities. It could support physicians when prescreening cases by allowing for invasive examinations to be performed only for cases involving problems with automatic recognition or listening and for professional analyses of other clinical examination results that reveal doubts about the presence of pathologies.

1 July 2025 in JASA Express Letters
Verified
Journal Article
Open access
TTTzu-Pei Tsai
Tzu-Pei Tsai
TBTessa Bent
Tessa Bent
MHMalachi Henry
Malachi Henry
This study investigated how variation in vocal timbre (e.g., neutral and twangy) influences intelligibility and listening effort in traffic-shaped noises. Word identification and subjective listening effort ratings were used to assess timbre effects. Intelligibility was significantly higher in twangy speech than in neutral speech. Twangy timbre had lower listening effort compared to neutral timbre for female speech only. The twang advantage might arise from higher perceived loudness, higher F1 and F2, and less breathy and hoarse voice quality. The result suggests that vocal timbre adjustments could be a strategy talkers employ to improve intelligibility in noise.