Joint prosodic and spectral modeling for robust speaker verification

Keyphrases

• Spectral Features 100%
• Spectral Modeling 100%
• Prosodic Modeling 100%
• Robust Speaker Verification 100%
• Domain Fusion 100%
• Fusion Approach 50%
• Handset 50%
• Equal Error Rate 50%
• Feature-based 50%
• Detection Task 50%
• Modeling Framework 50%
• Ambient Noise 50%
• Hierarchical Structure 50%
• Speaker Change Detection 50%
• State Dependence 50%
• Transition State 50%
• Speaker Recognition Evaluation 50%
• Speech Prosody 50%
• State Transition Probability 50%
• Ergodic HMM 50%
• NIST SRE 50%
• Dependent Observations 50%
• Status Transition 50%
• Ergodic Model 50%
• Single-speaker 50%

Computer Science

• Spectral Feature 100%
• Speaker Verification 100%
• Experimental Result 50%
• Equal Error Rate 50%
• And-States 50%
• Speaker Recognition 50%
• State Transition 50%
• Observation Distribution 50%
• Modeling Framework 50%