Overlap in Frequency
In the beginning, it was assumed that the frequency components of various sources do not overlap one another. In fact, there is, of course, always a chance that one or more components do overlap. This possibility is not constant but depends on the coincidental frequency content of the sources and the frequency resolution in the digital processing. The consequences depend on the amplitude ratios and the width of the phase window. The source delivering the strongest contribution at that moment has the strongest influence on the net phase difference of the total of both overlapping components.
If the contribution from the desired source is strongest, the component will be considered desirable. The contribution from the interfering source will then be unjustly passed, based on the phase difference. This will manifest itself in a lower suppression of the interfering source.
If the contribution of the interfering source is strongest, the component will be considered interference. The contribution of the desired source will then be unjustly suppressed based on the phase difference. If we make the window very small, we can hear this with speech in combination with noise as a loss in the high tones. The high tones are normally lower in amplitude. Weaker components of a desired source are generally less relevant. With the combination of several overlapping frequency components and whereby the contribution of the interfering source to these components is stronger than those of the desired source, we will miss much of the desired source. Depending on the coincidental overlap, fragments occur whereby the desired signal fails partially or completely.
Last update: September 24, 2006