BDDM: Bilateral Denoising Diffusion Models for Fast and High-Quality Speech Synthesis.
Max W. Y. LamJun WangDan SuDong YuPublished in: ICLR (2022)
Keyphrases
- speech synthesis
- diffusion models
- denoising
- high quality
- speech recognition
- diffusion model
- image denoising
- information diffusion
- text to speech
- social networks
- vocal tract
- prosodic features
- noisy images
- image processing
- natural images
- viral marketing
- high resolution
- influence maximization
- speech signal
- anisotropic diffusion
- noise reduction
- objective function
- bayesian networks