T-GSA: Transformer with Gaussian-weighted self-attention for speech enhancement.

Jaeyoung Kim Mostafa El-Khamy Jungwon Lee

Published in: CoRR (2019)

Keyphrases

speech enhancement
noisy environments
noise reduction
linear prediction
speech signal
single channel
computational complexity
signal to noise ratio
focus of attention
wiener filter
multiscale
maximum likelihood
visual attention
vocal tract
smoothing algorithm
information retrieval