Blind stochastic feature transformation for channel robust speaker verification

Keyphrases

• Stochastic Characteristics 100%
• Channel Compensation 100%
• Transformation Approach 100%
• Feature Transformation 100%
• Robust Speaker Verification 100%
• Result-oriented 50%
• Composite Model 50%
• Statistical Model 50%
• Feature-based 50%
• Compensation Algorithm 50%
• Telephone-based 50%
• Speaker Recognition System 50%
• Transformation Parameters 50%
• Handset 50%
• Cepstral Mean Subtraction 50%
• Background Modeling 50%
• Channel Characteristics 50%
• Channel Variation 50%
• Detection Cost 50%
• Stochastic Matching 50%
• Matching Type 50%
• NIST Evaluation 50%
• Speaker Model 50%
• Equal Error Rate 50%

Engineering

• Experimental Result 100%
• Priori Knowledge 100%
• Channel Variation 100%
• Telephone 100%
• Equal Error Rate 100%
• Statistical Model 100%

Computer Science

• Speaker Verification 100%
• Verification System 50%
• Equal Error Rate 50%
• Transformation Parameter 50%
• Composite Model 50%
• Telephone 50%
• cepstral 50%
• Experimental Result 50%
• Priori Knowledge 50%
• Background Model 50%
• Statistical Model 50%