Bilinear map of filter-bank outputs for DNN-based speech recognition.
Tetsuji OgawaKenshiro UedaKouichi KatsuradaTetsunori KobayashiTsuneo NittaPublished in: INTERSPEECH (2015)
Keyphrases
- speech recognition
- filter bank
- image coding
- subband
- hidden markov models
- multiresolution
- multiscale
- automatic speech recognition
- language model
- wavelet packet
- pattern recognition
- perfect reconstruction
- speech recognizer
- signal processing
- speech recognition technology
- wavelet transform
- speech signal
- wavelet filters
- computationally efficient
- frequency domain
- speaker identification
- speech synthesis
- speech recognition systems
- neural network
- singular value decomposition
- speaker independent
- training process
- probabilistic model
- spectral analysis
- denoising