2D-to-2D Mask Estimation for Speech Enhancement Based on Fully Convolutional Neural Network.
Yanhui TuJun DuChin-Hui LeePublished in: ICASSP (2020)
Keyphrases
- convolutional neural network
- speech enhancement
- face detection
- noise reduction
- noisy environments
- speech signal
- signal to noise ratio
- speech recognition
- background noise
- wiener filter
- single channel
- image processing
- non stationary
- language model
- frequency domain
- machine learning
- multiresolution
- linear prediction
- vocal tract
- moving objects