用于混响背景语音分离的结构稀疏模型(Strutured sparisty model)方法-To further tackle the ambiguity of the reflection ratios, we propose a novel formulation of the reverberation model and estimate the absorption coefficients through a convex optimization exploiting