Parallel WaveNet: Fast High-Fidelity Speech Synthesis

[TanUkkii007]

link: https://deepmind.com/blog/high-fidelity-speech-synthesis-wavenet/

paper: https://deepmind.com/documents/131/Distilling_WaveNet.pdf

referenced from:

• https://twitter.com/heiga_zen/status/933435493089513473
• FFTNET: A REAL-TIME SPEAKER-DEPENDENT NEURAL VOCODER #105

[TanUkkii007]

Further readings

perceptual loss

• A Neural Algorithm of Artistic Style https://arxiv.org/abs/1508.06576
• Perceptual Losses for Real-Time Style Transfer and Super-Resolution https://arxiv.org/abs/1603.08155

discretized mixture of logistics distribution

• PixelCNN++: Improving the PixelCNN with Discretized Logistic Mixture Likelihood and Other Modifications #32

Inverse Autoregressive Flow

• Improving Variational Inference with Inverse Autoregressive Flow https://arxiv.org/abs/1606.04934

related methods to "probability density distillation"

• Non-Autoregressive Neural Machine Translation #33

[TanUkkii007]

Done.