Target Speech Enhancement Method based on degenerate unmixing and estimation technique

摘要

PURPOSE: A target speech enhancement method based on degenerated unmixing is provided to apply a real application without estimating time delay coefficient without the number of sound source signal. CONSTITUTION: A first channel signal and a second channel signal are converted into time-frequency function(200). A histogram about a parameter is generated by estimating the parameter of a first channel signal and a second channel signal(210). An initial value of the parameter is set up about interest sound source through a histogram(220).