From Discrete Tokens to High-Fidelity Audio Using Multi-Band Diffusion

Keyphrases

• Audio 100%
• Discrete Tokens 100%
• Mel-spectrogram 66%
• Low Dose Rate 33%
• Generative Techniques 33%
• Modeling Approach 33%
• Bit Rate 33%
• Diffusion Model 33%
• GitHub Projects 33%
• Discrete Representation 33%
• Diffusion-based 33%
• Deep Generative Models 33%
• Low Sampling Rate 33%
• Environmental Sounds 33%
• Perceptual Quality 33%
• Mel Frequency Cepstral Coefficients 33%
• Use-diffusion 33%
• Evaluation Codes 33%
• Compressed Representation 33%
• Vocoder 33%
• Audio Modality 33%

Computer Science

• Generative Technique 100%
• Diffusion Model 100%
• Deep Generative Model 100%
• Perceptual Quality 100%
• Low Sampling Rate 100%
• Mel Frequency Cepstral Coefficient 100%

Engineering

• Spectrogram 100%
• Generative Model 50%
• Sampling Rate 50%
• Bit Rate 50%
• Diffusion Model 50%
• Perceptual Quality 50%