Scalable and Efficient Neural Speech Coding: A Hybrid Design

Keyphrases

• Hybrid Design 100%
• Neural Waves 100%
• Neural Speech Coding 100%
• Codec 87%
• Bit Rate 37%
• High Bit Rate 25%
• Linear Predictive Coding 25%
• Autoencoding 25%
• Residual Coding 25%
• Residual Learning 25%
• Crossed Module 25%
• Compact Model 12%
• Process Optimization 12%
• Low Dose Rate 12%
• Concatenated 12%
• Downscaling 12%
• Coding Artifacts 12%
• System Yield 12%
• Superior Performance 12%
• Coding System 12%
• Scalable Architecture 12%
• TIMIT 12%
• Decoder 12%
• Entropy Coding 12%
• Speech Coder 12%
• Code Quality 12%
• System Training 12%
• Convolutional Neural Network 12%
• Depthwise Separable Convolution 12%
• Code Module 12%
• Coding Problems 12%
• Trainable 12%
• Bitrate Control 12%
• Decoding Complexity 12%
• Speech Coding 12%
• Test Utterance 12%
• WaveNet 12%
• Autoencoder 12%
• Vocoder 12%
• Waveform Coding 12%
• Transparent Performance 12%
• Opus 12%
• Quantization Coding 12%
• Reconstruction Loss 12%
• Speech Codec 12%
• Speech Compression 12%
• AMR-WB 12%
• Gated Residual Networks 12%
• Wideband Speech Coding 12%
• Coding Concepts 12%

Computer Science

• Quantization (Signal Processing) 100%
• Predictive Coding 100%
• Coding Artifact 50%
• Superior Performance 50%
• Component Model 50%
• Training System 50%
• Process Optimization 50%
• Convolutional Neural Network 50%
• Decoding Complexity 50%
• Autoencoder 50%
• Waveform Coding 50%
• Residual Neural Network 50%
• Coding Quality 50%
• Speech Compression 50%
• Depthwise Separable Convolution 50%

Engineering

• Quantization (Signal Processing) 100%
• Residual Coding 100%
• Tasks 50%
• Coding Artifact 50%
• Feedforward 50%
• Component Model 50%
• Decoding Complexity 50%
• Autoencoder 50%
• Convolutional Neural Network 50%
• Coding Quality 50%