Leveraging gain normalization for sub-band temporal features in noise-robust speech recognition.
Hao-Teng FanJeih-Weih HungPublished in: FSKD (2012)
Keyphrases
- speech recognition
- noisy environments
- subband
- low frequency
- high frequency
- hidden markov models
- speech signal
- discrete wavelet transform
- automatic speech recognition
- wavelet transform
- speech synthesis
- language model
- pattern recognition
- background noise
- frequency domain
- speech enhancement
- speech recognition systems
- noisy speech
- multiresolution
- filter bank
- wavelet coefficients
- speech recognizer
- high pass
- speaker identification
- wavelet domain
- image compression
- speech recognition technology
- low pass
- bit rate
- wavelet decomposition
- feature vectors
- computer vision
- additive noise
- speaker independent
- noise reduction
- speaker dependent
- neural network
- denoising
- word recognition
- frequency band
- signal to noise ratio