A reduced complexity MFCC-based deep neural network approach for speech enhancement.
Ryan RazaniHanwook ChungYazid AttabiBenoît ChampagnePublished in: ISSPIT (2017)
Keyphrases
- reduced complexity
- speech enhancement
- speech signal
- neural network
- noisy speech
- noisy environments
- speech recognition
- vector quantization
- noise reduction
- single channel
- signal to noise ratio
- background noise
- linear prediction
- speaker identification
- hidden markov models
- multi channel
- pattern recognition
- wiener filter
- motion estimation algorithm
- automatic speech recognition
- vocal tract
- image coder
- non stationary
- smoothing algorithm
- additive noise
- image quality
- multiresolution