Employing median filtering to enhance the complex-valued acoustic spectrograms in modulation domain for noise-robust speech recognition.
Hsin-Ju HsiehBerlin ChenJeih-weih HungPublished in: ISCSLP (2016)
Keyphrases
- speech recognition
- median filtering
- complex valued
- noisy environments
- speech recognition systems
- speech signal
- median filter
- noisy speech
- speaker independent
- post processing
- hidden markov models
- real valued
- speech recognizer
- automatic speech recognition
- additive noise
- jpeg compression
- acoustic models
- language model
- pattern recognition
- gaussian noise
- speaker identification
- recurrent neural networks
- vector field
- blind source separation
- image analysis
- noise level
- image enhancement
- neural network
- motion estimation
- noise reduction