Monaural Speech Enhancement Based on Spectrogram Decomposition for Convolutional Neural Network-sensitive Feature Extraction.
Hao ShiLongbiao WangSheng LiJianwu DangTatsuya KawaharaPublished in: INTERSPEECH (2022)
Keyphrases
- convolutional neural network
- speech enhancement
- single channel
- speech signal
- feature extraction
- frequency domain
- multi channel
- noisy environments
- face detection
- speech recognition
- vocal tract
- linear prediction
- face recognition
- signal to noise ratio
- noise reduction
- background noise
- feature vectors
- automatic speech recognition
- additive noise
- principal component analysis
- wiener filter
- image processing
- pattern classification
- non stationary
- spectral analysis
- independent component analysis
- wavelet transform
- spatial domain
- neural network
- signal processing
- support vector machine
- feature space
- pattern recognition
- smoothing algorithm
- machine learning
- extracted features
- hidden markov models
- multiresolution
- sound source
- object recognition
- video sequences
- feature selection