SNR-Invariant Multitask Deep Neural Networks for Robust Speaker Verification

Keyphrases

• Signal-to-noise Ratio 100%
• Robust Speaker Verification 100%
• I-vector 100%
• Multi-task Deep Neural Network 100%
• Deep Neural Network 80%
• Hierarchical Regression 40%
• Utterance 40%
• Speaker Dependent 40%
• Deep Learning Architectures 40%
• Auxiliary Task 40%
• Denoising 20%
• Ratio-dependent 20%
• Speaker Verification 20%
• Noisy Environment 20%
• Background Noise 20%
• Verification Performance 20%
• Multi-task Learning 20%
• Speaker Information 20%
• Mixture of PLDA 20%
• Multi-conditions 20%
• Multi-task DNN 20%
• Cluster Mean 20%
• Clustering Phenomena 20%
• Lower Error Rates 20%

Computer Science

• Deep Neural Network 100%
• Speaker Verification 100%
• Noise-to-Signal Ratio 100%
• Neural Network Architecture 18%
• Multitask Learning 9%
• Experimental Result 9%
• Noisy Environment 9%
• Background Noise 9%