Personalized speech enhancement through self-supervised data augmentation and purification

Keyphrases

• Data Purification 100%
• Speech Enhancement 100%
• Data Augmentation 100%
• Pseudo-source 100%
• Enhancement Model 40%
• Predictive Models 20%
• Learning Problems 20%
• Noisy Data 20%
• Freedom of Speech 20%
• Limited Access 20%
• Noise Free 20%
• Speech Recordings 20%
• Privacy-preserving 20%
• Clear Speech 20%
• Speech in Noise 20%
• Self-supervised Learning 20%
• Privacy Constraint 20%
• Supervised Approach 20%
• Target Speaker 20%
• Personalized Model 20%
• Denoising Model 20%
• Speaker Embedding 20%
• Source Frame 20%
• Model Personalization 20%

Computer Science

• Speech Enhancement 100%
• Data Augmentation 100%
• de-noising 25%
• Straightforward Approach 25%
• Learning Problem 25%
• Privacy Preserving 25%
• Self-Supervised Learning 25%