VACE-WPE: Virtual Acoustic Channel Expansion Based on Neural Networks for Weighted Prediction Error-Based Speech Dereverberation.
Joon-Young YangJoon-Hyuk ChangPublished in: IEEE ACM Trans. Audio Speech Lang. Process. (2022)
Keyphrases
- prediction error
- neural network
- reversible watermarking
- linear prediction
- speech sounds
- acoustic features
- motion compensated
- speech recognition systems
- inter frame
- speech recognition
- multi channel
- bit rate
- automatic speech recognition
- linear predictors
- back propagation
- motion vectors
- emotional speech
- prosodic features
- vocal tract
- acoustic signal
- formant frequencies
- speech signal
- speaker independent
- speaker verification
- motion estimation
- vowel phonemes
- block matching
- hidden markov models
- computer vision