Blind stochastic feature transformation for speaker verification over cellular networks

Keyphrases

• Test Utterance 100%
• Matched Condition 50%
• Transformation Approach 50%
• Channel Variation 50%
• Detection Cost 50%
• Amount of Training 50%
• Inter-speaker Variation 50%
• Gaussianization 50%
• Speaker Recognition 50%
• Gender-dependent 50%
• Speaker Gender 50%

Engineering

• Equal Error Rate 100%
• Statistical Difference 100%
• Channel Variation 100%
• Speaker Recognition 100%

Computer Science

• Background Model 100%
• System Condition 50%
• Speaker Recognition 50%