SpecDiff-GAN: A Spectrally-Shaped Noise Diffusion GAN for Speech and Music Synthesis.
Teysir BaouebHaocheng LiuMathieu FontaineJonathan Le RouxGaël RichardPublished in: CoRR (2024)
Keyphrases
- structuring elements
- audio signals
- median filter
- anisotropic diffusion
- noise level
- speech music discrimination
- noisy environments
- random noise
- audio features
- noise reduction
- signal to noise ratio
- noisy data
- speech enhancement
- speech signal
- binary images
- digital audio
- music information retrieval
- additive noise
- image noise
- gray scale
- speech recognition
- audio visual
- music recommendation
- remote sensing
- audio recordings
- noise model
- noise removal
- gaussian noise
- background noise
- spoken language
- broadcast news
- mathematical morphology
- program synthesis
- missing data