Reconstructing clean speech from noisy MFCC vectors.
Ben MilnerJonathan DarchIbrahim AlmajaiPublished in: INTERSPEECH (2009)
Keyphrases
- noisy speech
- speech recognition
- noisy environments
- speech signal
- speech enhancement
- hidden markov models
- background noise
- speaker identification
- feature vectors
- automatic speech recognition
- noisy observations
- noise reduction
- signal to noise ratio
- mel frequency cepstral coefficients
- speaker verification
- speaker recognition
- vocal tract
- language model
- spectral subtraction
- vector space
- pattern recognition
- speech recognition systems
- speech synthesis
- feature extraction
- acoustic features
- linear prediction
- spectral analysis
- cepstral coefficients
- bayesian networks
- missing data
- recognition engine
- speaker diarization
- speech recognizer
- discrete tomography
- noisy data
- gaussian mixture model