Prosodic Modeling for Speaker Recognition Based on Sub-Band Energy Temporal Trajectories.
André Gustavo AdamiPublished in: ICASSP (1) (2005)
Keyphrases
- speaker recognition
- subband
- speech recognition
- wavelet transform
- gaussian mixture model
- filter bank
- discrete wavelet transform
- feature vectors
- multiresolution
- image compression
- high frequency
- wavelet coefficients
- bit rate
- speaker verification
- vector quantization
- speaker identification
- speech signal
- low pass
- multiscale
- pattern recognition
- model selection
- probabilistic neural network
- neural network
- pattern classification
- noise reduction
- multi modal
- machine learning