Specmix: a mixed sample data augmentation method for training with time-frequency domain features

Keyphrases

• Acoustic Scene Classification 33%
• Audio 33%
• Audio Data Augmentation 33%
• Audio Sample 33%
• Audio Scene Classification 33%
• Augmentation Approach 33%
• Augmentation Techniques 66%
• Classification Performance 33%
• Classification Refinement 66%
• Data Augmentation 100%
• Image Classification 33%
• Mixed Samples 100%
• Neural Network Architecture 33%
• Sound Event Classification 66%
• Spectral Correlation 33%
• Speech Enhancement 66%
• Time-frequency Domain Features 100%
• Time-frequency Mask 33%

Engineering

• Classification Performance 33%
• Data Sample 33%
• Domain Feature 100%
• Frequency Domain 100%
• Image Classification 33%
• Neural Network Architecture 33%
• Sample Data 100%
• Spectral Correlation 33%
• Speech Enhancement 66%

Computer Science

• Classification Performance 33%
• Data Augmentation 100%
• Domain Feature 100%
• Frequency Domain 100%
• Image Classification 33%
• Neural Network Architecture 33%
• Spectral Correlation 33%
• Speech Enhancement 66%