Learning speaker aware offsets for speaker adaptation of neural networks

Keyphrases

• Speaker Adaptation 100%
• Neural Network Adaptation 100%
• Speaker-aware 100%
• Adaptation 80%
• Long Short-term Memory 80%
• Storage Layer 60%
• Senone 60%
• Auxiliary Network 60%
• Word Error Rate 40%
• Fully Connected Layer 40%
• Adaptation Studies 20%
• Mean Squared Error 20%
• Feature Space 20%
• Normalization Method 20%
• Classification Accuracy 20%
• Rate Reduction 20%
• Adversarial Model 20%
• Training Framework 20%
• Adversarial Training 20%
• Maximum Likelihood Linear Regression 20%
• Layer Normalization 20%
• Memory Activation 20%

Computer Science

• Neural Network 100%
• Long Short-Term Memory Network 100%
• Word Error Rate 66%
• maximum-likelihood 33%
• Feature Space 33%
• Classification Accuracy 33%
• Adversarial Machine Learning 33%
• Adversarial Model 33%
• Long Short-Term Memory Networks 33%
• Attached Network 33%