FSK and compander levels are adjusted.
Eliminate offsets between subsequent speech chunks. This is done by
high-pass filter. An offset is not passed to the filter.
Do correct audio processing chain:
time compress -> compressor -> scrambler / pre-emphasis -> TX
RX -> de-scrambler / de-emphasis -> expander -> time expand
This is done by simulating two cells with different polarities.
On first message from mobile station, the cell with the correct
polarity is used, the other is disabled.
On RX side the polarity is not selected, since FSK decoder detects
polarity for every frame.
Andreas Eversberg