Speaker normalized spectral subband parameters for noise robust speech recognition.
Satoru TsugeToshiaki FukudaHarald SingerPublished in: ICASSP (1999)
Keyphrases
- speech recognition
- subband
- noisy environments
- filter bank
- speaker verification
- wavelet transform
- low frequency
- speech signal
- linear prediction
- automatic speech recognition
- image compression
- speaker identification
- bit rate
- speaker adaptation
- wavelet coefficients
- high frequency
- hidden markov models
- language model
- wavelet decomposition
- multiresolution
- feature vectors
- frequency domain
- speech synthesis
- pattern recognition
- cepstral coefficients
- discrete wavelet transform
- speech recognizer
- speaker dependent
- speaker independent
- speaker recognition
- wavelet domain
- additive noise
- speech recognition systems
- maximum likelihood
- noise reduction
- signal to noise ratio
- neural network
- speaker diarization
- multiscale
- computer vision
- spectral analysis
- power spectrum
- image coding
- non stationary
- vocal tract
- image segmentation