Until now, we have talked about total sound pressure levels, or the combined levels in all frequencies. Most sounds fluctuate in sound pressure levels and frequencies over time. These fluctuations could be plotted in a three-dimensional graph, with time and frequencies on the two horizontal axes and sound pressure level on the vertical axis. However, such a graph would be complicated and confusing. To simplify the display, a moment in time may be examined two-dimensionally, with frequency on the horizontal axis and sound pressure level on the vertical axis. Such a representation is called a "frequency spectrum". Much useful information can be extracted from a frequency spectrum. The peaks show what frequencies are dominant. From a noise control perspective, reducing the peaks in the dominant frequencies will have the greatest effect on reducing the overall noise levels. Also, high peaks in certain frequencies suggest a tonal characteristic of the sound. Tonal sounds have concentrated sound pressure levels in one or several frequencies and tend to be more annoying to human ears than sound pressure levels that are spread out over many frequencies.