envelope rms implementation review

I am trying to implement the moving RMS by Matlab.

x = randn(50, 1);
xRMS = sqrt(movmean(x.^2, 21));
xRMSref = envelope(x, 21, 'rms');
plot(xRMSref,'DisplayName','xRMSref');hold on;plot(xRMS,'DisplayName','xRMS');hold off;
legend()

Why my estimation differs from Matlab's? What actually implemented by Matlab? The algorithm description is not delailed and movrms function is locked...

[Nim Pim]

Hi. I too found that moving RMS function of dsp tool box not giving the correct values. I tried it on a small data set and the manually calculated values were different. So confusing....

Answer 1

[Chunru]

"envelope" removes the mean first before doing movrms and it restore the mean offset in the end.

If you nake sure the signal is 0-mean, then the results would be quite same.

x = randn(50, 1);
x = x -mean(x);
xRMS = sqrt(movmean(x.^2, 21));
xRMSref = envelope(x, 21, 'rms');
plot(xRMSref,'DisplayName','xRMSref');
hold on;
plot(xRMS,'DisplayName','xRMS');hold off;
legend()