Time-regularized Linear Prediction for Noise-robust Extraction of the Spectral Envelope of Speech.
Manu AiraksinenLauri JuvelaOkko RäsänenPaavo AlkuPublished in: INTERSPEECH (2018)
Keyphrases
- linear prediction
- speech enhancement
- speech signal
- noisy environments
- image coding
- lossless compression
- prediction error
- spectral analysis
- speech recognition
- subband
- vocal tract
- information extraction
- automatic speech recognition
- autoregressive
- noise reduction
- signal to noise ratio
- image compression
- speaker identification
- least squares
- wavelet coefficients
- non stationary
- edge detection
- speaker recognition
- multiresolution
- cepstral coefficients