Far Field Speech Enhancement at Low SNR in Presence of Nonstationary Noise Based on Spectral Masking and MVDR Beamforming.
Sergei AstapovAleksandr LavrentyevEvgeniy ShuranovPublished in: SPECOM (2018)
Keyphrases
- speech enhancement
- non stationary
- low snr
- speech signal
- linear prediction
- noisy environments
- spectral analysis
- noise reduction
- autoregressive
- signal to noise ratio
- noise model
- single channel
- vocal tract
- background noise
- low signal to noise ratio
- additive noise
- wiener filter
- noise level
- automatic speech recognition
- lossless compression
- random fields
- image coding
- smoothing algorithm
- speech recognition
- blind source separation
- image restoration
- signal processing