Estimating direct-to-reverberant ratio mapped from power spectral density using deep neural network.
Yusuke HiokaKenta NiwaPublished in: ICASSP (2016)
Keyphrases
- power spectral density
- neural network
- spectral analysis
- fourier transform
- speech signal
- power spectrum
- frequency band
- prediction error
- artificial neural networks
- blind separation
- neural network model
- back propagation
- pattern recognition
- speech recognition
- subband
- autoregressive
- activation function
- video sequences
- optimal filter
- face recognition