Figure 2From: An efficient solution to sparse linear prediction analysis of speech The weighting function A frame of a voiced sound along with the detected GCI locations and the constructed weighting function (with σ = 50 and κ = 1 ). Back to article page