A Noise Prediction and Time-Domain Subtraction Approach to Deep Neural Network Based Speech Enhancement.
Babafemi O. OdelowoDavid V. AndersonPublished in: ICMLA (2017)
Keyphrases
- speech enhancement
- linear prediction
- noisy environments
- noise reduction
- signal to noise ratio
- single channel
- prediction error
- speech signal
- background noise
- image coding
- wiener filter
- neural network
- additive noise
- frequency domain
- noisy speech
- smoothing algorithm
- vocal tract
- edge preserving
- multi channel
- image restoration
- lossless compression
- random variables
- maximum likelihood