Filterbank-energy estimation using mixture and Markov models for recognition of noisy speech.
Adoram ErellMitchel WeintraubPublished in: IEEE Trans. Speech Audio Process. (1993)
Keyphrases
- markov models
- noisy speech
- hidden markov models
- filter bank
- markov model
- automatic speech recognition systems
- speech recognition
- subband
- image coding
- maximum entropy
- background noise
- higher order
- multiresolution
- spectral analysis
- conditional random fields
- noisy environments
- speech signal
- multiscale
- computationally efficient
- mixture model
- speech enhancement
- image processing
- frequency domain
- automatic speech recognition
- image quality
- wavelet transform