Time-Domain Loss Modulation Based on Overlap Ratio for Monaural Conversational Speaker Separation.

Hassan Taherian DeLiang Wang

Published in: ICASSP (2021)

Keyphrases

multi modal
frequency domain
speaker verification
speech recognition
automatic speech recognition
neural network
dimensionality reduction
standard deviation
error propagation
speaker recognition
contextual knowledge
information loss
audio visual
conversational agent