Speech Enhancement Based on Reducing the Detail Portion of Speech Spectrograms in Modulation Domain via DiscreteWavelet Transform.
Shih-Kuang LeeSyu-Siang WangYu TsaoJeih-weih HungPublished in: ISCSLP (2018)
Keyphrases
- speech enhancement
- speech signal
- single channel
- noisy environments
- noise reduction
- vocal tract
- multi channel
- signal to noise ratio
- noisy speech
- speech recognition
- linear prediction
- background noise
- hidden markov models
- non stationary
- smoothing algorithm
- wiener filter
- speaker identification
- automatic speech recognition
- multiscale
- spectral analysis
- additive noise
- prior information
- independent component analysis
- super resolution
- edge detection
- pattern recognition