Speech enhancement based on Rayleigh mixture modeling of speech spectral amplitude distributions.
Jan S. ErkelensJesper JensenRichard HeusdensPublished in: EUSIPCO (2007)
Keyphrases
- speech enhancement
- mixture modeling
- linear prediction
- speech signal
- noisy environments
- dirichlet distribution
- noise reduction
- gaussian distribution
- signal to noise ratio
- density estimation
- vocal tract
- single channel
- spectral analysis
- mixture model
- image coding
- noisy speech
- em algorithm
- speech recognition
- smoothing algorithm
- probability distribution
- prediction error
- speech synthesis
- wiener filter
- lossless compression
- automatic speech recognition
- background noise
- power law
- expectation maximization
- wavelet transform
- multi channel
- random variables
- gaussian mixture model
- discrete data
- image restoration
- spectral features
- sar images
- maximum likelihood