Improvements to harmonic model for extracting better speech features in clinical applications.
Computer Speech & Language(2018)
摘要
•Improvements on harmonic model leads to accurate estimation of voiced segments and pitch frequency in presence of background noise as well as clinical speech including inherent noises.•Time-varying property of the proposed model allows more accurate estimation of amplitude variations (shimmer) in a short-term waveform.•Acoustic features extracted using the proposed model are useful in clinical applications and experimental results on two clinical tasks show their advantages compare to standard acoustic features.
更多查看译文
关键词
Pitch tracking,Voice activity detection,Modified harmonic model
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络