Unvoiced speech segregation based on CASA and spectral subtraction.
Ke HuDeLiang WangPublished in: INTERSPEECH (2010)
Keyphrases
- speech signal
- spectral subtraction
- speech recognition
- automatic speech recognition
- speech quality
- linear prediction
- speaker identification
- noisy speech
- vocal tract
- background noise
- noisy environments
- hidden markov models
- speech enhancement
- spectral analysis
- non stationary
- noisy images
- optimal solution
- hearing aids
- speech synthesis
- neural network
- sound source
- pattern recognition
- noise level
- image coding
- optimization algorithm
- language model
- training data