Try something like this —
x = 1:20;
y = randn(size(x));
xi = linspace(min(x), max(x), numel(x)*10);
yi = interp1(x, y, xi);
Lv = yi > 0;
zv = zeros(size(xi));
figure
plot(x, y)
hold on
patch([xi flip(xi)], [zv flip(yi.*Lv)], 'k', 'FaceAlpha',0.75)
hold off
grid
Experiment with the ‘Noisydata.txt’ signal.
.
