Environment adaptation for robust speaker verification by cascading maximum likelihood linear regression and reinforced learning

Keyphrases

• Handset 100%
• Robust Speaker Verification 100%
• Environment Adaptation 100%
• Reinforcement Learning 100%
• Maximum Likelihood Linear Regression 100%
• Regression Learning 100%
• Speaker Model 40%
• Background Modeling 30%
• Feature Transformation 20%
• Result-oriented 10%
• Utterance 10%
• Speech Signal 10%
• Stochastic Characteristics 10%
• Transformed Model 10%
• Speaker Verification 10%
• Telephone Network 10%
• Acoustic Distortion 10%
• Model Synthesis 10%
• Learning Transformation 10%
• Distorted Speech 10%
• Speech Data 10%
• Distortion Degree 10%

Computer Science

• Speaker Verification 100%
• maximum-likelihood 100%
• Background Model 100%
• Synthesis Model 33%
• Experimental Result 33%
• Telephone Network 33%