So, I stated all this in my comments, but, although I have to answer this.
I set setInitialDelay () to 100. Then I populated 100 sample size buffers and read 2 points from it. A glance counted from point 0 when sound entered the system. Live audio (which the user hears) has always been 100 steps back.
This works well for my requirements. Just make sure that all array values ββare initialized, otherwise you will hear some terrible noise coming out of your speakers.
source share