An investigation of the reconstruction capacity of stacked convolutional autoencoders for log-mel-spectrograms

Engineering

• Spectrogram 100%
• Autoencoder 100%
• Metrics 66%
• Harmonics 66%
• Similarities 33%
• Initial Design 33%
• Regularization 33%
• Audio Signal Processing 33%
• Generative Model 33%
• Deep Learning Method 33%

Computer Science

• Audio Compression 66%
• Neural Network 33%
• Regularization 33%
• Deep Generative Model 33%
• Level Representation 33%
• Evaluation Metric 33%
• Unsupervised Manner 33%
• Audio Signal Processing 33%
• Deep Learning Method 33%

Chemical Engineering

• Deep Learning Method 100%
• Neural Network 100%