Two-Step Sound Source Separation: Training on Learned Latent Targets

Keyphrases

• Scale Invariant 100%
• Latent Space 100%
• Signal-to-Distortion Ratio 100%
• Sound Source Separation 100%
• Neural Network 50%
• Proposed Methodology 50%
• Large Classes 50%
• Separation System 50%
• Two-stage Training 50%
• Loss Function 50%
• Separation Performance 50%
• Oracle 50%
• Source Separation 50%
• Training Procedure 50%
• Deep Neural Network 50%
• Separation Experiment 50%
• Sound Separation 50%

Mathematics

• Neural Network 100%
• Time Domain 100%
• Loss Function 100%
• Sound Source 100%
• Deep Neural Network 100%
• Training Procedure 100%

Computer Science

• Source Separation 100%
• Neural Network 50%
• Deep Neural Network 50%
• Separation Performance 50%