Marginal energy density over the low frequency range as a feature for voiced/non-voiced detection in noisy speech signals.
Pooja JainRam Bilas PachoriPublished in: J. Frankl. Inst. (2013)
Keyphrases
- speech signal
- low frequency
- noisy environments
- high frequency
- speech recognition
- automatic speech recognition
- frequency domain
- speaker identification
- wavelet transform
- subband
- vocal tract
- non stationary
- high frequency components
- wavelet coefficients
- hidden markov models
- discrete wavelet transform
- speech enhancement
- frequency band
- high resolution
- low and high frequency
- bayesian networks
- automatic speech recognition systems
- feature vectors
- pattern recognition
- image quality
- denoising
- multiresolution
- fusion rules
- sound signals