Throughout this tutorial, you use and reuse the same filter with complex coefficients. This filter has 32 coefficients (or taps) and no symmetry characteristics.
{ -82, -253},{ 0, -204},{ 11, -35},{ -198, 273},
{ -642, 467},{ -1026, 333},{ -927, 0},{ -226, -73},
{ 643, 467},{ 984, 1355},{ 550, 1691},{ 0, 647},
{ 538, -1656},{ 2860, -3936},{ 6313, -4587},{ 9113, -2961},
{ 9582, 0},{ 7421, 2411},{ 3936, 2860},{ 1023, 1409},
{ -200, -615},{ 0, -1778},{ 517, -1592},{ 467, -643},
{ -192, 140},{ -882, 287},{ -1079, 0},{ -755, -245},
{ -273, -198},{ 22, 30},{ 63, 194},{ 0, 266}
The output of this filter has a much higher amplitude than the input. You must apply a scaling factor of 2^15 to get back to the normalized data. In the debugging phase, when only impulses are given to the filter, you can reduce the scaling factor to 1. This lets you verify that the output looks like the impulse response of the filter.