FNeural Speech Enhancement with Very Low Algorithmic Latency and Complexity via Integrated full- and sub-band Modeling.
Zhong-Qiu WangSamuele CornellShukjae ChoiYounglo LeeByeong-Yeol KimShinji WatanabePublished in: ICASSP (2023)
Keyphrases
- subband
- speech enhancement
- filter bank
- discrete wavelet transform
- wavelet transform
- high frequency
- multiresolution
- wavelet coefficients
- feature vectors
- image compression
- linear prediction
- bit rate
- computational complexity
- wavelet decomposition
- frequency domain
- frequency band
- noise reduction
- high resolution
- noisy environments
- vocal tract
- smoothing algorithm
- pattern recognition
- single channel
- wiener filter
- signal to noise ratio
- multiscale
- signal processing