Outlier Detection and Removal for HMM-Based Speech Synthesis with an Insufficient Speech Database
スポンサーリンク
概要
- 論文の詳細を見る
Decision tree-based clustering and parameter estimation are essential steps in the training part of an HMM-based speech synthesis system. These two steps are usually performed based on the maximum likelihood (ML) criterion. However, one of the drawbacks of the ML criterion is that it is sensitive to outliers which usually result in quality degradation of the synthesized speech. In this letter, we propose an approach to detect and remove outliers for HMM-based speech synthesis. Experimental results show that the proposed approach can improve the synthetic speech, particularly when the available training speech database is insufficient.
著者
-
Kim Nam
School Of Electrical And Computer Engineering Chungbuk National Univ.
-
HONG Doo
School of Electrical Engineering and the Institute of New Media and Communications, Seoul National University
-
SUNG June
School of Electrical Engineering and the Institute of New Media and Communications, Seoul National University
-
OH Kyung
School of Electrical Engineering and the Institute of New Media and Communications, Seoul National University
-
KIM Nam
School of Architecture and Architectural Engineering, Korea University of Technology and Education
関連論文
- Two-Dimensional Electrophoretic Analysis of Radio Frequency Radiation-Exposed MCF7 Breast Cancer Cells
- Feature Compensation with Model-Based Estimation for Noise Masking(Speech and Hearing)
- Computationally Efficient Cepstral Domain Feature Compensation
- On Detecting Target Acoustic Signals Based on Non-negative Matrix Factorization
- Improved Frame Mode Selection for AMR-WB+ Based on Decision Tree
- Estimation of Phone Mismatch Penalty Matrices for Two-Stage Keyword Spotting
- Implementation of HMM-Based Human Activity Recognition Using Single Triaxial Accelerometer
- Speech Enhancement Based on Perceptually Comfortable Residual Noise(Multimedia Systems for Communications)
- Study of Prominence Detection Based on Various Phone-Specific Features
- Frame Splitting Scheme for Error-Robust Audio Streaming over Packet-Switching Networks
- Three-Dimensional Display System Based on Integral Imaging with Viewing Direction Control
- Depth Discrimination Enhanced Computational Integral Imaging Using Random Pattern Illumination
- Speech Enhancement Based on Data-Driven Residual Gain Estimation
- Analysis of the Cellular Stress Response in MCF10A Cells Exposed to Combined Radio Frequency Radiation
- Outlier Detection and Removal for HMM-Based Speech Synthesis with an Insufficient Speech Database
- Spectral Magnitude Adjustment for MCLT-Based Acoustic Data Transmission
- Statistical Approaches to Excitation Modeling in HMM-Based Speech Synthesis
- APPLICATION OF RELIABILITY -BASED SAFETY FACTORS TO MECHANISTIC-EMPIRICAL FLEXIBLE PAVEMENT DESIGN