SpecDiff-GAN: A Spectrally-Shaped Noise Diffusion GAN for Speech and Music Synthesis.
Teysir BaouebHaocheng LiuMathieu FontaineJonathan Le RouxGaël RichardPublished in: ICASSP (2024)
Keyphrases
- structuring elements
- audio signals
- noisy environments
- speech music discrimination
- median filter
- noise level
- speech recognition
- noise reduction
- audio features
- speech synthesis
- speech signal
- additive noise
- speech enhancement
- missing data
- audio visual
- audio recordings
- image noise
- text to speech
- facial animation
- random noise
- music information retrieval
- anisotropic diffusion
- binary images
- music score
- denoising
- language model
- tensor field
- gray scale
- pattern recognition
- noisy images
- signal to noise ratio