Hello,

There are cases where my NLMS algorithm let's echo slip through. When
I observe how the filter coefficients adapt, I see a lot of
fluctuations around the peak. Is that an indicator of non-linearities?
Such as: phone case vibrations/reflections picked up by the mic which
the algorithm has a hard time dealing with? It seems that the
fluctuations around the peak only happens when the phone loudspeaker
is turned up too much and my guess is that the acoustic signal makes
the microphone signal's relationship with the speaker signal more non-
linear than linear. Is that a correct observation?