The present invention provides a simple method to correct cross-talk, after the data have been generated. Adjacent signals are simply subtracted from the original observed signal with a repeating process. The data processing is stopped when a predefined condition is met. By this technique, cross-talk can be reduced from >5% to less than 0.1%. And as an additional advantage, this method provides a way to correct the cross-talk without the need to know which peaks are caused by the adjacent capillary signal.